From nurani at ripe.net Mon May 8 18:26:19 2000 From: nurani at ripe.net (Nurani Nimpuno) Date: Mon, 08 May 2000 18:26:19 +0200 Subject: IP assignment for virtual webhosting Message-ID: <200005081626.SAA24698@x7.ripe.net> Dear lir-wg, At the end of 1999 we raised the issue of IP Assignment to virtual web-hosting and we believe that it now is high time to conclude this discussion. We received very useful feedback from the community and we also brought the discussion further to the other RIRs, ARIN and APNIC. This has now also been discussed in their respective regions. In the RIPE community there was tremendously strong endorsement for name-based web-hosting although there were also several arguments against a policy making it mandatory to use name-based web-hosting. There were technical arguments put forward, naming examples where IP-based web-hosting is necessary in order to allow certain technologies or applications such as SSL for example, but there were also voices raised against the hassle of upgrading old hard- and software. The RIPE NCC is in favour of a policy making it mandatory to use name-based web-hosting when technically feasible. However, due to some of the voices raised against this, we believe an accurate conclusion at this moment would be to not change any policy at this point, but to further strongly discourage IP-based web-hosting. We do however wish to raise our concern regarding what we see as an inefficient usage of addresses in our limited address pool. We would therefore like to request the community to consider making it mandatory for NEW installations to use domain based web-hosting, with the exception of a set of agreed applications needing IP based web-hosting (eg. SSL). We wish to add this to the agenda at the upcoming RIPE meeting for further discussion and welcome any input the community may have on this matter. Kind regards, Nurani Nimpuno Registration Services Manager RIPE NCC From nurani at ripe.net Mon May 8 18:57:19 2000 From: nurani at ripe.net (Nurani Nimpuno) Date: Mon, 08 May 2000 18:57:19 +0200 Subject: Minimum allocation size Message-ID: <200005081657.SAA24977@x7.ripe.net> Dear all, The RIPE NCC has been monitoring the consumption rates of the initial /19 allocations given to new members over the last couple of years. The results clearly show that a significant amount of these allocations are far from fully used. The majority of these allocations are in fact not used more than 50%. Due to these low usage rates, the RIPE NCC is considering changing the initial allocation size from the current /19 to a /20, in order to accommodate this change in the usage patterns. We will however present the community with clear results and statistics in the lir-wg at the coming RIPE meeting in Budapest for further discussion regarding this proposed policy change. We are looking forward to hearing the community's input on this matter in a constructive and fruitful discussion next week. Kind regards, Nurani Nimpuno Registration Services Manager RIPE NCC From woeber at cc.univie.ac.at Mon May 8 20:29:54 2000 From: woeber at cc.univie.ac.at (Wilfried Woeber, UniVie/ACOnet) Date: Mon, 08 May 2000 20:29:54 +0200 Subject: IP assignment for virtual webhosting Message-ID: <009E9CB4.A21D38B8.9@cc.univie.ac.at> Dear Nurani! >Subj: IP assignment for virtual webhosting Before making up my mind I'd like to better understand what (in the following parapraph) >We do however wish to raise our concern regarding what we see as an >inefficient usage of addresses in our limited address pool. We would >therefore like to request the community to consider making it >mandatory for NEW installations to use domain based web-hosting, with >the exception of a set of agreed applications needing IP based >web-hosting (eg. SSL). "NEW installations" is in reality? I could come up with a couple of different interpretations, like: - an existing ISP or Web-Hotel can use the old technology, new sites are required to use the new technology, or - replacing or upgrading en existing web server (hardware only? software only? both?) requires the new technology to be deployed, or - .... But probably it's just because I don't know too much about web hosting in the first place... Wilfried. _________________________________:_____________________________________ Wilfried Woeber : e-mail: Woeber at CC.UniVie.ac.at UniVie Computer Center - ACOnet : Tel: +43 1 4277 - 140 33 Universitaetsstrasse 7 : Fax: +43 1 4277 - 9 140 A-1010 Vienna, Austria, Europe : RIPE-DB: WW144, PGP keyID 0xF0ACB369 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ From lucks at indigo.ie Mon May 8 18:36:24 2000 From: lucks at indigo.ie (Sascha Luck) Date: Mon, 8 May 2000 17:36:24 +0100 Subject: IP assignment for virtual webhosting In-Reply-To: <200005081626.SAA24698@x7.ripe.net>; from nurani@ripe.net on Mon, May 08, 2000 at 06:26:19PM +0200 References: <200005081626.SAA24698@x7.ripe.net> Message-ID: <20000508173624.M46404@indigo.ie> Thus spoke Nurani Nimpuno: > mandatory for NEW installations to use domain based web-hosting, with > the exception of a set of agreed applications needing IP based > web-hosting (eg. SSL). I'd go one step further and make it mandatory to use HTTP 1.1 based hosting for _upgraded_ installations. I don't know how to enforce that, however. Otherwise, I can live with that proposal. Best regards, s. -- Sascha Luck | Multimedia Infrastructure Group Eircom PLC | Internet House / Teach Idirline phone: +353-1-7010900 | 26-34 TempleBar / Barra an Teampaill mailto: lucks at indigo.ie | Dublin 2 / Baile Atha Cliath 2 From Robert.Kiessling at de.easynet.net Mon May 8 18:36:52 2000 From: Robert.Kiessling at de.easynet.net (Robert Kiessling) Date: Mon, 8 May 2000 18:36:52 +0200 (MEST) Subject: IP assignment for virtual webhosting In-Reply-To: <200005081626.SAA24698@x7.ripe.net> References: <200005081626.SAA24698@x7.ripe.net> Message-ID: <14614.60708.566564.461196@doncamillo.local.easynet.de> Nurani Nimpuno writes: > We would > therefore like to request the community to consider making it > mandatory for NEW installations to use domain based web-hosting, with > the exception of a set of agreed applications needing IP based > web-hosting (eg. SSL). So you would require servers like www.siemens.de to share an IP address with www.my-completely-unimportant-personal-site.de if there is no SSL or similar use? I would suggest to make exceptions for heavily used sites, too. Since the vast mayority of webservers does not fall into this category, this is no waste of address space. Robert From lucks at indigo.ie Mon May 8 19:18:41 2000 From: lucks at indigo.ie (Sascha Luck) Date: Mon, 8 May 2000 18:18:41 +0100 Subject: Minimum allocation size In-Reply-To: <200005081657.SAA24977@x7.ripe.net>; from nurani@ripe.net on Mon, May 08, 2000 at 06:57:19PM +0200 References: <200005081657.SAA24977@x7.ripe.net> Message-ID: <20000508181841.N46404@indigo.ie> Thus spoke Nurani Nimpuno: > Due to these low usage rates, the RIPE NCC is considering changing > the initial allocation size from the current /19 to a /20, in order > to > accommodate this change in the usage patterns. How is a new LIR going to announce their routes then? s. -- Sascha Luck | Multimedia Infrastructure Group Eircom PLC | Internet House / Teach Idirline phone: +353-1-7010900 | 26-34 TempleBar / Barra an Teampaill mailto: lucks at indigo.ie | Dublin 2 / Baile Atha Cliath 2 From matthew at planet.net.uk Tue May 9 10:46:07 2000 From: matthew at planet.net.uk (Matthew Robinson) Date: Tue, 9 May 2000 09:46:07 +0100 Subject: IP assignment for virtual webhosting Message-ID: <6BF1C330AF53D311BE5D00508B09081001E3FE3B@PLANET01> But surely www.big-importantco.com would have their own machine and wouldn't share with www.my-little-home-page.org ? We try to apply the rule that you should only have unique addresses for ssl servers that can't do IP-less or some other server protocol that can't support it. Many of our customers grumble at first but soon realise that they can add their own IP-less virtuals without need to involve our admin department. All that's needed is a little DNS and the whole job can be completed in a day! My two pence Cheers Matthew -----Original Message----- From: Robert Kiessling [mailto:Robert.Kiessling at de.easynet.net] Sent: 08 May 2000 17:37 To: Nurani Nimpuno Cc: lir-wg at ripe.net Subject: IP assignment for virtual webhosting Nurani Nimpuno writes: > We would > therefore like to request the community to consider making it > mandatory for NEW installations to use domain based web-hosting, with > the exception of a set of agreed applications needing IP based > web-hosting (eg. SSL). So you would require servers like www.siemens.de to share an IP address with www.my-completely-unimportant-personal-site.de if there is no SSL or similar use? I would suggest to make exceptions for heavily used sites, too. Since the vast mayority of webservers does not fall into this category, this is no waste of address space. Robert From darko.bulat at k2.net Tue May 9 11:01:19 2000 From: darko.bulat at k2.net (Darko Bulat) Date: Tue, 09 May 2000 11:01:19 +0200 Subject: IP assignment for virtual webhosting References: <200005081626.SAA24698@x7.ripe.net> <14614.60708.566564.461196@doncamillo.local.easynet.de> Message-ID: <3917D3DF.DC093162@k2.net> Robert Kiessling wrote: > > Nurani Nimpuno writes: > > We would > > therefore like to request the community to consider making it > > mandatory for NEW installations to use domain based web-hosting, with > > the exception of a set of agreed applications needing IP based > > web-hosting (eg. SSL). > > So you would require servers like www.siemens.de to share an IP > address with www.my-completely-unimportant-personal-site.de if there > is no SSL or similar use? If they share the same hardware (eg. physical box serving both WWW sites), then what's the problem with sharing the same IP? Also, WWW is (was?!) supposed to bring some level of equality, so for some people Siemens might be important, for someone else that "unimportant" site might be the most important site in the world. What we can measure is how much traffic does some WWW produce and in "domain name seb-hosting" what matters is how many requests for a site fail because of outdated (old) HTML browsers. In any case we can be grateful that protocol exists so we can use "domain based web-hosting" and we can responsibly contribute to conserving IP4 space. (eg. we stopped assigning separate IP numbers for virtual web-hosts months ago) Regards! Darko > I would suggest to make exceptions for heavily used sites, too. Since > the vast mayority of webservers does not fall into this category, this > is no waste of address space. > > Robert From stephenb at uk.uu.net Tue May 9 10:59:21 2000 From: stephenb at uk.uu.net (Stephen Burley) Date: Tue, 09 May 2000 08:59:21 +0000 Subject: Minimum allocation size References: <200005081657.SAA24977@x7.ripe.net> <20000508181841.N46404@indigo.ie> Message-ID: <3917D369.8F82BCEA@uk.uu.net> Sascha Luck wrote: > Thus spoke Nurani Nimpuno: > > Due to these low usage rates, the RIPE NCC is considering changing > > the initial allocation size from the current /19 to a /20, in order > > to > > accommodate this change in the usage patterns. > > How is a new LIR going to announce their routes then? This is a very good point, the only way round this is to have a mid way transit point so 2 companies would share an announcment of a /19 i can not see very many people agreeing to that. Maybe if 2 the registries have a common upstream provider the provider would agree to aggregate, but even then this causes problems when changing providers. Do you have any more concrete proposals than the basic statement? Regards, Stephen Burley UUNET EMEA Hostmaster > > > s. > > -- > Sascha Luck | Multimedia Infrastructure Group > Eircom PLC | Internet House / Teach Idirline > phone: +353-1-7010900 | 26-34 TempleBar / Barra an Teampaill > mailto: lucks at indigo.ie | Dublin 2 / Baile Atha Cliath 2 From matthew at planet.net.uk Tue May 9 11:11:27 2000 From: matthew at planet.net.uk (Matthew Robinson) Date: Tue, 9 May 2000 10:11:27 +0100 Subject: Minimum allocation size Message-ID: <6BF1C330AF53D311BE5D00508B09081001E3FE3E@PLANET01> How many providers filter less than a /19? Can we not ask them to allow /20's from 'new' allocations? -----Original Message----- From: Stephen Burley [mailto:stephenb at uk.uu.net] Sent: 09 May 2000 9:59 To: lucks at indigo.ie Cc: Nurani Nimpuno; lir-wg at ripe.net Subject: Re: Minimum allocation size This is a very good point, the only way round this is to have a mid way transit point so 2 companies would share an announcment of a /19 i can not see very many people agreeing to that. Maybe if 2 the registries have a common upstream provider the provider would agree to aggregate, but even then this causes problems when changing providers. Do you have any more concrete proposals than the basic statement? From b.fiser at cybertron.at Tue May 9 11:06:10 2000 From: b.fiser at cybertron.at (Bernhard Fiser) Date: Tue, 09 May 2000 11:06:10 +0200 Subject: IP assignment for virtual webhosting References: <200005081626.SAA24698@x7.ripe.net> <14614.60708.566564.461196@doncamillo.local.easynet.de> Message-ID: <3917D502.F11C3315@cybertron.at> Robert Kiessling wrote: > Nurani Nimpuno writes: > > We would > > therefore like to request the community to consider making it > > mandatory for NEW installations to use domain based web-hosting, with > > the exception of a set of agreed applications needing IP based > > web-hosting (eg. SSL). > > So you would require servers like www.siemens.de to share an IP > address with www.my-completely-unimportant-personal-site.de if there > is no SSL or similar use? > > I would suggest to make exceptions for heavily used sites, too. Since > the vast mayority of webservers does not fall into this category, this > is no waste of address space. > > Robert I agree. bh From joao at ripe.net Tue May 9 10:53:09 2000 From: joao at ripe.net (Joao Luis Silva Damas) Date: Tue, 9 May 2000 10:53:09 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: <009E9CB4.A21D38B8.9@cc.univie.ac.at> References: <009E9CB4.A21D38B8.9@cc.univie.ac.at> Message-ID: Dear all, maybe it would be clearer for everyone if instead of "new installations" we say "new (IP) allocations", because that's more along the lines we were thinking. The idea is not to force old installations to reinvent themselves due to a policy change, rather to make more efficient use of address space in the future, based on currently available technology. Also the list of exceptions is up for discussion, SSL was an initial kick-off example but it is ultimately the ISPs that have to agree on the list. We'll be happy to assemble the list and will propose all applications we see ourselves as needing IP based hosting, but it would be impossible for us to come up with a complete list of all the imaginative applications of technology that ISPs are using. Also I don't think such a list would have to be "set in stone" as, fortunately, the Internet is constantly evolving. We look forward to a discussion on the list and next week at the RIPE meeting. Joao Damas Head of External Services RIPE NCC From greg at askfol.com Tue May 9 11:03:13 2000 From: greg at askfol.com (Greg Lloyd Smith) Date: Tue, 9 May 2000 10:03:13 +0100 Subject: We cannot really stop anyone from becoming a LIR... -- RIPE Hostmaster Message-ID: The following is an extract from a series of communications on the subject of RIPE having recently authorised the addition of another LIR in the UK, which uses the same trading name of an existing member. It is our feeling that RIPE should not have allowed the addition of this LIR on the basis that the company in question trades using a name which could easily confuse the general public as to which company is which and that since our company was an existing member, and had previously been a LIR in another country as well, they should have decline the application of the LIR on the basis that their trading name could cause a conflict and confuse the general public, and would further cause damage to an existing member's company and reputation. Greg Lloyd Smith uk.firstnetonline [RIPE NCC Hostmaster writes] Good morning Greg, [uk.firstnetonline writes] "FirstNET Online PLC" writes: * Thank you for your continued discussion on this matter.... What you could * have done, was determined whether or not RIPE needed (yet) another LIR, and * base that decision on the same factors I was presented with in Greece... * That is.. Could they have obtained their IP addresses from their up line * connectivity provider, and was there another member using the name FirstNET * or something similar... These were the questions I was asked when I applied * to RIPE. I am very distressed by the fact that about a week ago there was * no other member of RIPE using our name and now there is and (most * important), there is no connection between our companies. [RIPE NCC Hostmaster writes] Yes we did all that. Its standard procedure as written in ripe-160. [uk.firstnetonline writes] * I would point out for the record that of the 20 companies I own two of them * (FirstNET Online Management Limited and FirstNET Online PLC), and that only * three of them are purporting to be ISP's and therefore the change of 20 of * them asking you for membership would be limited to the company about which * we are complaining and our own 2 companies. * * I would therefore ask that we be allowed to appeal to someone within the * RIPE organisation to have their membership revoked or repealed since it * should not have been granted in the first place. I was vetted and I suspect * that if there had been a company in Greece using the name Firstnet, you * would have rejected my application. Why was their application approved? [RIPE NCC Hostmaster writes] If your argument is that FirstNet Services should not have been allowed to become a Local IR because they partly use the same name as you then you have no agreement according to current policy. We cannot really stop anyone from becoming a LIR, only recommend them to go elsewhere for their address space. But if you feel strongly about it why don't you write to the LIR working group mailing list, , and state your case there. Policy is brought about here by the community, not by us. From henning.brauer at bsmail.de Tue May 9 10:48:22 2000 From: henning.brauer at bsmail.de (henning.brauer at bsmail.de) Date: Tue, 9 May 2000 10:48:22 +0200 Subject: IP assignment for virtual webhosting Message-ID: Hi all, there is one big problem i see: If you charge your customers for traffic, you must count the traffic, and the easiest and most reliable way to do this is on the firewall/router. Thats the way we do and therefore we need one IP per customer. Most customers have more than onme site and for that we are using HTTP 1.1 (NameVirtualHosts). As long as there is no Package for apache to cout the traffic !reliable! (the traffic counted via access logs is << real traffic ) I do not see a way to put all customers to a couple of IPs. Greetinmgs from Germany Henning Brauer Hostmaster BSWS ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 PLEASE USE EMAIL WHERE POSSIBLE Sascha Luck .ie> cc: lir-wg at ripe.net Sent by: Subject: Re: IP assignment for virtual webhosting owner-lir-wg@ ripe.net 08.05.00 18:36 Please respond to lucks Thus spoke Nurani Nimpuno: > mandatory for NEW installations to use domain based web-hosting, with > the exception of a set of agreed applications needing IP based > web-hosting (eg. SSL). I'd go one step further and make it mandatory to use HTTP 1.1 based hosting for _upgraded_ installations. I don't know how to enforce that, however. Otherwise, I can live with that proposal. Best regards, s. -- Sascha Luck | Multimedia Infrastructure Group Eircom PLC | Internet House / Teach Idirline phone: +353-1-7010900 | 26-34 TempleBar / Barra an Teampaill mailto: lucks at indigo.ie | Dublin 2 / Baile Atha Cliath 2 From nurani at ripe.net Tue May 9 11:37:51 2000 From: nurani at ripe.net (Nurani Nimpuno) Date: Tue, 09 May 2000 11:37:51 +0200 Subject: Minimum allocation size In-Reply-To: Your message of Tue, 09 May 2000 08:59:21 -0000. <3917D369.8F82BCEA@uk.uu.net> References: <3917D369.8F82BCEA@uk.uu.net> Message-ID: <200005090937.LAA18281@x7.ripe.net> Dear all, The purpose of my initial mail was simply to draw you attention to this coming discussion. We have a full set of data and statistics that we wish to present to the community in the lir-wg at the coming RIPE meeting in Budapest. I believe it is more useful to discuss the matter in detail once all facts have been presented to you. This is a policy change that has taken place in both the APNIC and ARIN region, where the RIRs notified their respective communities of this policy change. As a result of this, routing filters have been adjusted accordingly and /20s are currently being announced without any problem. We therefore are not greatly concerned about the practical implications of such a change. We do however want the community's input as to whether such a change is desirable in the RIPE region. As we will have representatives from both APNIC and ARIN present in the coming RIPE meeting, we propose to discuss this in more detail next week. Statistics and data from our region will be presented together with arguments both for and against such a change. Kind regards, Nurani Stephen Burley writes: * Sascha Luck wrote: * * > Thus spoke Nurani Nimpuno: * > > Due to these low usage rates, the RIPE NCC is considering changing * > > the initial allocation size from the current /19 to a /20, in order * > > to * > > accommodate this change in the usage patterns. * > * > How is a new LIR going to announce their routes then? * * This is a very good point, the only way round this is to have a mid way * transit point so 2 companies would share an announcment of a /19 i can * not see very many people agreeing to that. Maybe if 2 the registries * have a common upstream provider the provider would agree to aggregate, * but even then this causes problems when changing providers. Do you have * any more concrete proposals than the basic statement? * * Regards, * Stephen Burley * UUNET EMEA Hostmaster * * * * * > * > * > s. * > * > -- * > Sascha Luck | Multimedia Infrastructure Group * > Eircom PLC | Internet House / Teach Idirline * > phone: +353-1-7010900 | 26-34 TempleBar / Barra an Teampaill * > mailto: lucks at indigo.ie | Dublin 2 / Baile Atha Cliath 2 * * From Havard.Eidnes at runit.sintef.no Tue May 9 11:51:40 2000 From: Havard.Eidnes at runit.sintef.no (Havard.Eidnes at runit.sintef.no) Date: Tue, 09 May 2000 11:51:40 +0200 Subject: We cannot really stop anyone from becoming a LIR... -- RIPE Hostmaster In-Reply-To: Your message of "Tue, 9 May 2000 10:03:13 +0100" References: Message-ID: <20000509115140N.he@runit.sintef.no> > The following is an extract from a series of communications on the > subject of RIPE having recently authorised the addition of another > LIR in the UK, which uses the same trading name of an existing > member. It is our feeling that RIPE should not have allowed the > addition of this LIR on the basis that the company in question > trades using a name which could easily confuse the general public > as to which company is which and that since our company was an > existing member, and had previously been a LIR in another country > as well, they should have decline the application of the LIR on > the basis that their trading name could cause a conflict and > confuse the general public, and would further cause damage to an > existing member's company and reputation. Hm, this seems like a strange request. The point of contest here seems to be the trading name used by the company who is now a (newly established) LIR, not that organization's ability to become a LIR. I don't see how the RIPE NCC can be concerned about the use of a trading name by an applicant for LIR status, and the possible infringement on someone else's rights to that name or the potential for resulting market confusion. This issue should instead be brought before the apporpriate authority, which I assume would be some part of the legal system in the UK. Of course, I'm not a lawyer, and I even don't play one on the net, but thus just seemed too obvious... - H?vard From niek at knoware.nl Tue May 9 11:45:39 2000 From: niek at knoware.nl (Niek Rijnbout) Date: Tue, 09 May 2000 11:45:39 +0200 Subject: We cannot really stop anyone from becoming a LIR... -- RIPE Hostmaster In-Reply-To: Message-ID: <4.3.1.2.20000509114107.00be9780@bobo.knoware.nl> At 10:03 9-05-00 +0100, you wrote: >The following is an extract from a series of communications on the subject >of RIPE having recently authorised the addition of another LIR in the UK, >which uses the same trading name of an existing member. It is our feeling >that RIPE should not have allowed the addition of this LIR on the basis that >the company in question trades using a name which could easily confuse the >general public as to which company is which and that since our company was >an existing member, and had previously been a LIR in another country as >well, they should have decline the application of the LIR on the basis that >their trading name could cause a conflict and confuse the general public, >and would further cause damage to an existing member's company and >reputation. > > >Greg Lloyd Smith >uk.firstnetonline Greg, what general public are you talking about. There is no general public looking at the RIPE database to look for company names. If you are pissed that somebody founded a company with a similar name as one of your companies, and want to take action, there are other means to do that. I do not think it is right to try to use RIPE to make somebody elses live problematic. Niek Rijnbout. Knoware. From niilo.neuvo at saunalahti.fi Tue May 9 11:39:56 2000 From: niilo.neuvo at saunalahti.fi (Niilo Neuvo) Date: Tue, 09 May 2000 12:39:56 +0300 Subject: Minimum allocation size In-Reply-To: <3917D369.8F82BCEA@uk.uu.net> Message-ID: <200005090939.MAA22028@telakka.saunalahti.fi> Sascha Luck wrote: > Thus spoke Nurani Nimpuno: > > Due to these low usage rates, the RIPE NCC is considering changing > > the initial allocation size from the current /19 to a /20, in order > > to > > accommodate this change in the usage patterns. > > How is a new LIR going to announce their routes then? Does anyone have any estimates on how much more routes this change would cause? I would suspect that this would take several years before even a 10% increase would be seen. Since this does not mean that you have to break the current aggregations. We have found this /19 to be rather large. We are a merger of about 20 small ISPs and also several LIRs. Now we are in the process of either filling up several /19 (since returning them to RIPE was a bigger job) and renumbering ISPs that never became LIRs. Starting with a /20 and somehow getting all of the ISPs to become LIRs would be great. Low cost entry LIRs at /20 and some simple web-based method for allocating the space (get them hooked!). Once they fill their /20 they'd have to become real LIRs. Unfortunately I can't guarantee that this actually saves address space in the long run. Since these mergers that we have seen here in Finland are bound to happen everywhere else once the market penetration is high enough. -- Niilo Neuvo >>>^<<< niilo.neuvo at saunalahti.fi CTO /$\ +358 (0) 50 5611042 (mobile) SAUNALAHTI Oyj | | +358 (0) 50 85611042 (fax) .88.744/7.88.744/7.88.744/7.88.oOOOo.88.744/7.88.744/7.88.744/7.88.74 From mike.norris at heanet.ie Tue May 9 12:16:44 2000 From: mike.norris at heanet.ie (mike.norris at heanet.ie) Date: Tue, 9 May 2000 11:16:44 +0100 Subject: We cannot really stop anyone from becoming a LIR... -- RIPE Hostmaster In-Reply-To: <3F2D1A940FB8D1118A1F00609783612938E8C1@ntserver.heanet.ie> Message-ID: <3F2D1A940FB8D1118A1F006097836129357057@ntserver.heanet.ie> > > The following is an extract from a series of communications on the > > subject of RIPE having recently authorised the addition of another > > LIR in the UK, which uses the same trading name of an existing > > member. It is our feeling that RIPE should not have allowed the > > addition of this LIR on the basis that the company in question > > trades using a name which could easily confuse the general public > > as to which company is which and that since our company was an > > existing member, and had previously been a LIR in another country > > as well, they should have decline the application of the LIR on > > the basis that their trading name could cause a conflict and > > confuse the general public, and would further cause damage to an > > existing member's company and reputation. > > Hm, this seems like a strange request. > > The point of contest here seems to be the trading name used by the > company who is now a (newly established) LIR, not that organization's > ability to become a LIR. I don't see how the RIPE NCC can be > concerned about the use of a trading name by an applicant for LIR > status, and the possible infringement on someone else's rights to that > name or the potential for resulting market confusion. This issue > should instead be brought before the apporpriate authority, which I > assume would be some part of the legal system in the UK. > > Of course, I'm not a lawyer, and I even don't play one on the net, but > thus just seemed too obvious... I agree with H?vard. RIPE NCC has no authority to adjudicate on entitlement to trading names; that's a matter for the parties concerned and perhaps the Companies Office in the UK. Regards. Mike From Peter.Baumann at commcare.ch Tue May 9 12:11:12 2000 From: Peter.Baumann at commcare.ch (Peter Baumann) Date: Tue, 09 May 2000 12:11:12 +0200 Subject: Minimum allocation size (Not Reachable) Message-ID: Hello I'm away from 9.5.00 until 12.5.00. Please try to reach me from the 15.5.00. Peter Baumann Commcare AG From hacar at ere.com.tr Tue May 9 12:11:13 2000 From: hacar at ere.com.tr (Hakki ACAR) Date: Tue, 9 May 2000 13:11:13 +0300 Subject: RIPE backlog References: <3.0.5.32.20000509124706.007d4710@max.ibm.net.il> Message-ID: <005001bfb99e$e7acdc20$1f01a8c0@ere.com.tr> I am waiting for the same subject since May 2 to get response from hm. I think they are very busy regarding these subjects ----- Original Message ----- From: "Hank Nussbacher" To: Sent: 09 Mayis 2000 Sali 13:47 Subject: RIPE backlog > I submitted an ASN request to hostmaster in April 27 and got back: > > >Please note that currently our working wait queue is long and you > >might have to wait up to a week for an initial response from a hostmaster. > >We are sorry for this delay and we are doing our best to bring the response > >times back to normal as soon as possible. > > We are now well past one week. I think the RIPE NCC should at the very > least be sending out notices to its customers in the hold queue that they > are still backlogged. > > Are others waiting just as long? > > -Hank > From leigh at insnet.net Tue May 9 14:33:45 2000 From: leigh at insnet.net (Leigh Porter) Date: Tue, 09 May 2000 13:33:45 +0100 Subject: RIPE backlog References: <3.0.5.32.20000509124706.007d4710@max.ibm.net.il> Message-ID: <391805A9.BF18608C@insnet.net> Joao Luis Silva Damas wrote: > > Dear Hank, > > yes, we currently have a longer than usual (and definitely longer > than desirable) delay in responding to member's requests for IP > addresses and AS numbers. Is there any way to quantify the delays being experianced? Perhaps an average time to initial hostmaster reply to a request or something that could be automatically mailed to requestees along with perhaps that requests postition in the queue. Just so that people get an idea of how long the queue is and when to expect a reply. Then the discussion here could be moved from "there is a delay how long is it" to something else :) Thanks, Leigh Porter INS From dnada2 at dgics.mod.uk Tue May 9 12:58:57 2000 From: dnada2 at dgics.mod.uk (Nick Naylor) Date: Tue, 9 May 2000 11:58:57 +0100 Subject: Fw: Minimum allocation size (Not Reachable) Message-ID: <000c01bfb9a5$9383e420$1ef05092@dgics.mod.uk> Misdirected email - I think! ----- Original Message ----- From: Peter Baumann To: Sent: Tuesday, May 09, 2000 11:11 AM Subject: Re: Minimum allocation size (Not Reachable) Hello I'm away from 9.5.00 until 12.5.00. Please try to reach me from the 15.5.00. Peter Baumann Commcare AG From hank at att.net.il Tue May 9 13:21:41 2000 From: hank at att.net.il (Hank Nussbacher) Date: Tue, 09 May 2000 13:21:41 +0200 Subject: We cannot really stop anyone from becoming a LIR... -- RIPE Hostmaster In-Reply-To: <3F2D1A940FB8D1118A1F006097836129357057@ntserver.heanet.ie> References: <3F2D1A940FB8D1118A1F00609783612938E8C1@ntserver.heanet.ie> Message-ID: <3.0.5.32.20000509132141.007d0210@max.ibm.net.il> At 11:16 09/05/00 +0100, mike.norris at heanet.ie wrote: > >> > The following is an extract from a series of communications on the >> > subject of RIPE having recently authorised the addition of another >> > LIR in the UK, which uses the same trading name of an existing >> > member. It is our feeling that RIPE should not have allowed the >> > addition of this LIR on the basis that the company in question >> > trades using a name which could easily confuse the general public >> > as to which company is which and that since our company was an >> > existing member, and had previously been a LIR in another country >> > as well, they should have decline the application of the LIR on >> > the basis that their trading name could cause a conflict and >> > confuse the general public, and would further cause damage to an >> > existing member's company and reputation. >> >> Hm, this seems like a strange request. >> >> The point of contest here seems to be the trading name used by the >> company who is now a (newly established) LIR, not that organization's >> ability to become a LIR. I don't see how the RIPE NCC can be >> concerned about the use of a trading name by an applicant for LIR >> status, and the possible infringement on someone else's rights to that >> name or the potential for resulting market confusion. This issue >> should instead be brought before the apporpriate authority, which I >> assume would be some part of the legal system in the UK. >> >> Of course, I'm not a lawyer, and I even don't play one on the net, but >> thus just seemed too obvious... > >I agree with H?vard. RIPE NCC has no authority to adjudicate on >entitlement to trading names; that's a matter for the parties concerned >and perhaps the Companies Office in the UK. Ditto. Not RIPE NCC's place to be playing trademark judge. -Hank > >Regards. > >Mike > > From dmorton at cybernet-ag.net Tue May 9 12:39:24 2000 From: dmorton at cybernet-ag.net (Dave Morton) Date: Tue, 09 May 2000 12:39:24 +0200 Subject: We cannot really stop anyone from becoming a LIR... -- RIPE Hostmaster References: <3F2D1A940FB8D1118A1F006097836129357057@ntserver.heanet.ie> Message-ID: <3917EADC.77556A90@cybernet-ag.net> mike.norris at heanet.ie wrote: > > > > The following is an extract from a series of communications on the > > > subject of RIPE having recently authorised the addition of another > > > LIR in the UK, which uses the same trading name of an existing > > > member. It is our feeling that RIPE should not have allowed the > > > addition of this LIR on the basis that the company in question > > > trades using a name which could easily confuse the general public > > > as to which company is which and that since our company was an > > > existing member, and had previously been a LIR in another country > > > as well, they should have decline the application of the LIR on > > > the basis that their trading name could cause a conflict and > > > confuse the general public, and would further cause damage to an > > > existing member's company and reputation. > > > > Hm, this seems like a strange request. > > > > The point of contest here seems to be the trading name used by the > > company who is now a (newly established) LIR, not that organization's > > ability to become a LIR. I don't see how the RIPE NCC can be > > concerned about the use of a trading name by an applicant for LIR > > status, and the possible infringement on someone else's rights to that > > name or the potential for resulting market confusion. This issue > > should instead be brought before the apporpriate authority, which I > > assume would be some part of the legal system in the UK. > > > > Of course, I'm not a lawyer, and I even don't play one on the net, but > > thus just seemed too obvious... > > I agree with Hevard. RIPE NCC has no authority to adjudicate on > entitlement to trading names; that's a matter for the parties concerned > and perhaps the Companies Office in the UK. In general I agree. But perhaps RIPE NCC should or has already asked their lawyers (sigh!) about the potential legal implications (if any) of not checking for an existing LIR using the same name (and in the same country). RIPE NCC should't have to bother with such disputes but there may or may not be liability considerations - sez he also not a lawyer but knowing how they always manage to make a mountain out of molehill. > Regards. > > Mike Regards as always, Dave "It's hard to make a comeback if you haven't been anywhere" - Anon "...and the news from Guadalajara where the temperature is 96 degrees, is that Falcao is warming up." From hank at att.net.il Tue May 9 12:47:06 2000 From: hank at att.net.il (Hank Nussbacher) Date: Tue, 09 May 2000 12:47:06 +0200 Subject: RIPE backlog Message-ID: <3.0.5.32.20000509124706.007d4710@max.ibm.net.il> I submitted an ASN request to hostmaster in April 27 and got back: >Please note that currently our working wait queue is long and you >might have to wait up to a week for an initial response from a hostmaster. >We are sorry for this delay and we are doing our best to bring the response >times back to normal as soon as possible. We are now well past one week. I think the RIPE NCC should at the very least be sending out notices to its customers in the hold queue that they are still backlogged. Are others waiting just as long? -Hank From joao at ripe.net Tue May 9 12:40:53 2000 From: joao at ripe.net (Joao Luis Silva Damas) Date: Tue, 9 May 2000 12:40:53 +0200 Subject: RIPE backlog In-Reply-To: <391805A9.BF18608C@insnet.net> References: <3.0.5.32.20000509124706.007d4710@max.ibm.net.il> <391805A9.BF18608C@insnet.net> Message-ID: At 13:33 +0100 5/9/00, Leigh Porter wrote: >Joao Luis Silva Damas wrote: >> >> Dear Hank, >> >> yes, we currently have a longer than usual (and definitely longer >> than desirable) delay in responding to member's requests for IP >> addresses and AS numbers. > >Is there any way to quantify the delays being experianced? Perhaps >an average time to initial hostmaster reply to a request or something >that could be automatically mailed to requestees along with perhaps >that requests postition in the queue. Being worked on as we "speak". > >Just so that people get an idea of how long the queue is and when to >expect a reply. > >Then the discussion here could be moved from "there is a delay how long >is it" to something else :) I agree. Currently we have about 8 business days of wait for first contact. Joao Damas RIPE NCC >Thanks, >Leigh Porter >INS From joao at ripe.net Tue May 9 12:12:19 2000 From: joao at ripe.net (Joao Luis Silva Damas) Date: Tue, 9 May 2000 12:12:19 +0200 Subject: RIPE backlog In-Reply-To: <3.0.5.32.20000509124706.007d4710@max.ibm.net.il> References: <3.0.5.32.20000509124706.007d4710@max.ibm.net.il> Message-ID: Dear Hank, yes, we currently have a longer than usual (and definitely longer than desirable) delay in responding to member's requests for IP addresses and AS numbers. I agree with you that it would have been better to notify the members who are suffering this condition. I would like to apologize for the overlook. Sometimes one gets so focused on reducing the wait time, other duties suffer. Again apologies as we are working to fix the current situation. regards, Joao Damas Head of External Services RIPE NCC At 12:47 +0200 5/9/00, Hank Nussbacher wrote: >I submitted an ASN request to hostmaster in April 27 and got back: > >>Please note that currently our working wait queue is long and you >>might have to wait up to a week for an initial response from a hostmaster. >>We are sorry for this delay and we are doing our best to bring the response >>times back to normal as soon as possible. > >We are now well past one week. I think the RIPE NCC should at the very >least be sending out notices to its customers in the hold queue that they >are still backlogged. > >Are others waiting just as long? > >-Hank From tim at colt.net Tue May 9 11:41:10 2000 From: tim at colt.net (Tim Franklin) Date: Tue, 9 May 2000 10:41:10 +0100 Subject: We cannot really stop anyone from becoming a LIR... -- RIPE Hostmaster In-Reply-To: Message-ID: <001701bfb99a$b5194ea0$4dd7fea9@travelling-mack.noc.colt.net> > > The following is an extract from a series of communications on the subject > of RIPE having recently authorised the addition of another LIR in the UK, > which uses the same trading name of an existing member. It is our feeling > that RIPE should not have allowed the addition of this LIR on the > basis that > the company in question trades using a name which could easily confuse the > general public as to which company is which and that since our company was > an existing member, and had previously been a LIR in another country as > well, they should have decline the application of the LIR on the > basis that > their trading name could cause a conflict and confuse the general public, > and would further cause damage to an existing member's company and > reputation. > To the best of my understanding, the general public have *very* little knowledge of LIRs. They go to ISPs, buy services and obtain IP addresses as part of that service. They won't, on the whole, fill out a 141 or usually have any idea that such a thing even exists. Certainly the only place I ever see an LIR name used is in the 'X-NCC-RegID' header, which I'm sure isn't seen by customers. How are similar LIR names going to confuse the general public when they never see them? Or are you suggesting that because another company has a similar name to yours, RIPE should attempt to cripple their entry into the ISP market? RIPE should only be interested in technical factors; if a company will be making assignments of blocks of IP addresses to their customers in sufficient quantities, and is capable of following all the allocation / documentation procedures, surely they have a strong case for becoming an LIR? Regards, Tim. -- Tim Franklin Email: tim at colt.net Project Engineer Phone: +44 20 7390 7848 COLT Internet Fax: +44 20 7863 5876 From lists at bridgenet.ch Tue May 9 14:54:58 2000 From: lists at bridgenet.ch (philip bridge) Date: Tue, 09 May 2000 13:54:58 +0100 Subject: Waiting to start RIPE-174 Arbitration In-Reply-To: References: <391805A9.BF18608C@insnet.net> <3.0.5.32.20000509124706.007d4710@max.ibm.net.il> <391805A9.BF18608C@insnet.net> Message-ID: <4.2.0.58.20000509134844.00989ec0@mail.nextra.ch> We want to start a an Arbitration process based on RIPE-174 to resolve a dispute with another LIR. I have already sent the request to initiate the process via the RIPE hostmaster route...which seems to be the intended route. The thing is, it seems that to *start* the process, we have to go through the same wait Q as normal hostmaster requests...adding another 1-2 weeks to a process that can already take up to 12 weeks. Is there a quicker way to initiate the process? Thanks Phil ______________________________________________________________ Philip Bridge www.bridgenet.ch ______________________________________________________________ It might look like I'm doing nothing, but at the cellular level I'm really quite busy... From adrian.pauling at bt.com Tue May 9 13:38:54 2000 From: adrian.pauling at bt.com (adrian.pauling at bt.com) Date: Tue, 9 May 2000 12:38:54 +0100 Subject: We cannot really stop anyone from becoming a LIR... -- RIPE H ostmaster Message-ID: <27EDC2145E42D211AD9600606DD5EC1D03BD3181@mbrpb1nt02.mww.bt.com> Nominet ( the .uk Domain Registrars ) now retain legal advisors due to the trademark / trading name implications with Domain Names. I'm not sure about other ccTLD Registrars. If RIPE-NCC were to perform trademark/trading name checks, they would have to employ a Trademark Attorney or such specialist legal expert. This would take time to recruit and would increase costs and extend the LIR set-up time. However, it is probably inappropriate for RIPE-NCC to do this. Whilst I sympathise with your position, I'd recommend no legal significance be attached to an LIR's name, and would suggest that your concerns about confusion between the uk.firstnetonlinexxxx is addressed via consultation with Trademark Attorneys (they do exist :-)). Regards, Adrian F Pauling :-)NEL2C Internet Protocol Manager acd Information Systems Engineering Technical Architecture AFP1-RIPE / AFP-ARIN / AFP25-InterNIC * adrian.pauling at bt.com * +44 19 2685 1992 / +44 78 0290 4877 British Telecommunications plc Registered Office 81 Newgate Street London EC1A 7AJ Registered in England no 1800000 > -----Original Message----- > From: Greg Lloyd Smith [SMTP:greg at askfol.com] > Sent: 09 May 2000 10:03 > To: lir-wg > Subject: We cannot really stop anyone from becoming a LIR... -- RIPE > Hostmaster > > The following is an extract from a series of communications on the subject > of RIPE having recently authorised the addition of another LIR in the UK, > which uses the same trading name of an existing member. It is our feeling > that RIPE should not have allowed the addition of this LIR on the basis > that > the company in question trades using a name which could easily confuse the > general public as to which company is which and that since our company was > an existing member, and had previously been a LIR in another country as > well, they should have decline the application of the LIR on the basis > that > their trading name could cause a conflict and confuse the general public, > and would further cause damage to an existing member's company and > reputation. > > > Greg Lloyd Smith > uk.firstnetonline > > > [RIPE NCC Hostmaster writes] > Good morning Greg, > > > [uk.firstnetonline writes] > "FirstNET Online PLC" writes: > * Thank you for your continued discussion on this matter.... What you > could > * have done, was determined whether or not RIPE needed (yet) another LIR, > and > * base that decision on the same factors I was presented with in > Greece... > * That is.. Could they have obtained their IP addresses from their up > line > * connectivity provider, and was there another member using the name > FirstNET > * or something similar... These were the questions I was asked when I > applied > * to RIPE. I am very distressed by the fact that about a week ago there > was > * no other member of RIPE using our name and now there is and (most > * important), there is no connection between our companies. > > [RIPE NCC Hostmaster writes] > Yes we did all that. Its standard procedure as written in ripe-160. > > [uk.firstnetonline writes] > * I would point out for the record that of the 20 companies I own two of > them > * (FirstNET Online Management Limited and FirstNET Online PLC), and that > only > * three of them are purporting to be ISP's and therefore the change of 20 > of > * them asking you for membership would be limited to the company about > which > * we are complaining and our own 2 companies. > * > * I would therefore ask that we be allowed to appeal to someone within > the > * RIPE organisation to have their membership revoked or repealed since it > * should not have been granted in the first place. I was vetted and I > suspect > * that if there had been a company in Greece using the name Firstnet, you > * would have rejected my application. Why was their application > approved? > > [RIPE NCC Hostmaster writes] > If your argument is that FirstNet Services should not have been > allowed to become a Local IR because they partly use the same name as > you then you have no agreement according to current policy. We cannot > really stop anyone from becoming a LIR, only recommend them to go > elsewhere for their address space. > > But if you feel strongly about it why don't you write to the LIR > working group mailing list, , and state your case > there. Policy is brought about here by the community, not by us. > > > > From nurani at ripe.net Tue May 9 19:05:09 2000 From: nurani at ripe.net (Nurani Nimpuno) Date: Tue, 09 May 2000 19:05:09 +0200 Subject: ripe-147 syntax checker Message-ID: <200005091705.TAA16846@x7.ripe.net> Dear all, The RIPE NCC is happy to announce a new web-based tool available to our members for AS number requests. This tool performs format- and syntax checking of AS Number requests as described in document ripe-147: "European Autonomous System Number Application Form and Supporting Notes": http://www.ripe.net/ripe/docs/ripe-147.html The tool can be found at the following URL: http://www.ripe.net/cgi-bin/web147cgi The purpose of AS number request syntax checker is to enable smoother and more accurate processing of AS number requests, minimising lack of information or incorrectly completed ripe-147 forms. Please note that this tool is only for preliminary format- and syntax checking of AS number requests. In order to obtain an AS number, a request needs to be sent to the the RIPE NCC hostmasters at , where it will be processed. We hope our members find this tool useful and we welcome any feedback and comments on this tool. Kind regards, Nurani Nimpuno Registration Services Manager RIPE NCC From briand at teleglobe.net Wed May 10 03:29:45 2000 From: briand at teleglobe.net (Brian Dickson) Date: Tue, 09 May 2000 21:29:45 -0400 Subject: IP assignment for virtual webhosting In-Reply-To: matthew's message of "Tue, 09 May 2000 09:46:07 +0100". <6BF1C330AF53D311BE5D00508B09081001E3FE3B@PLANET01> Message-ID: <200005100129.VAA23828@spare.de.teleglobe.net> I hate jumping into such a contentious issue, but the very fact that it is contentious, and that it will potentially impact both my employer, Teleglobe, and its customers, I have an obligation to point out several technical issues. I'll start with some recent threads: > But surely www.big-importantco.com would have their own machine and wouldn't > share with www.my-little-home-page.org ? The issue with non-IP based virtuals, is the difficulty and complexity, and more important, time-sensitivity involved, in *relocating* between servers. DNS-based virtuals are impacted by caching resolvers, which are ubiquitous. Lowering timers has risk too, especially if load spikes occur, eg for longer than timer values. If everyone is aware of the "slashdot" effect, then you can appreciate wanting to move such a "slashdotted" site to a separate server (or server farm), in a timeframe less than the DNS normal-TTL-value times. Not doing so would have a serious negative impact on 100 little-domains, who may have SLA's with you. Replace "slashdot" with "IPO" or "new software release", or any of a dozen other usage-spike events; each customer has their own traffic trigger, and the ISP won't necessarily have adequate warning. Three days is way too long; even three hours can result in bad press, angry customers, and cancelled contracts. > Nurani Nimpuno writes: > > We would > > therefore like to request the community to consider making it > > mandatory for NEW installations to use domain based web-hosting, with > > the exception of a set of agreed applications needing IP based > > web-hosting (eg. SSL). I think there are a number of issues concerning the definition of "NEW". Perhaps the only one to which most will agree is a square-one startup, with no pre-existing hardware, software, or address space. New address space, new servers, or even a new upstream, resulting in a mandatory upgrade places undue burden on any affected company, not the least of which is management systems for a new hosting system. And even there, there is another issue - the digital divide, the technical haves and have-nots. The first and third world. While there may not be as many RIPE members with customers in that category, the fact that *any* do means the policy must, IMHO, be sensitive to such issues. (Any RIPE members with offices in certain countries, eg USA, may be under legal obligation to do so, even if their head office, and/or customer, are not in said country.) There are many parts of the world, which would include parts of Europe which were formerly Iron Curtain countries, where the general expertese, available hardware and software, and language barriers, combined with newly-established free-market economies, put high demand for technology in the hands of the technical newbies, who are soooo far back on the learning curve that there is zero chance of them being proficient enough to deploy and configure DNS-based virtual domains as their first hosting service in a correct, or timely, fashion. I believe, that while it is admirable to encourage, strongly, those who can, to do DNS-based IP, that sufficient arguments in several areas run counter to mandatory policies for this technology. The vast majority of RIPE members are, I believe, commercial entities. Any policy which makes it necessary to turn away customers, would send the wrong message to the market in general. We all benefit from fair and open competition. If we permit "grandfathering" of IP-based virtuals, then to not permit new applicants to do so is ever-so-slightly unfair. ISPs need address space; this is a commercial issue for them. While we may hold ignorance and incompetence in contempt, (and I do, personally), I don't believe it is sufficient justification for the commercial impact of denying IP address space which is otherwise justified (via IP-based virtual hosting.) My two cents worth. -- Brian Dickson, Email: briand at teleglobe.net Director, Backbone Engineering Tel : +1 703 755 2056 Teleglobe Communications Corp., Fax : +1 703 755 2648 Rm 3214, 11480 Commerce Park Drive, Cell : +1 703 851 9053 Reston, Virginia, USA, 20191 http://www.teleglobe.com From michael.hallgren at teleglobe.com Wed May 10 10:29:24 2000 From: michael.hallgren at teleglobe.com (Hallgren, Michael) Date: Wed, 10 May 2000 09:29:24 +0100 Subject: IP assignment for virtual webhosting Message-ID: <04BCD7586FEDD3119C0B00A0C9E4605F6F99ED@uklozms02.Teleglobe.CA> Robert Kiessling wrote: > Nurani Nimpuno writes: > > We would > > therefore like to request the community to consider making it > > mandatory for NEW installations to use domain based web-hosting, with > > the exception of a set of agreed applications needing IP based > > web-hosting (eg. SSL). > > So you would require servers like www.siemens.de to share an IP > address with www.my-completely-unimportant-personal-site.de if there > is no SSL or similar use? Isn't it a bit unlikely that such a "power-site" would be sitting on a shared server (as avirtual Web) ? mh > > I would suggest to make exceptions for heavily used sites, too. Since > the vast mayority of webservers does not fall into this category, this > is no waste of address space. > > Robert I agree. bh From phk at critter.freebsd.dk Wed May 10 10:50:33 2000 From: phk at critter.freebsd.dk (Poul-Henning Kamp) Date: Wed, 10 May 2000 10:50:33 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: Your message of "Wed, 10 May 2000 09:29:24 BST." <04BCD7586FEDD3119C0B00A0C9E4605F6F99ED@uklozms02.Teleglobe.CA> Message-ID: <30837.957948633@critter.freebsd.dk> In message <04BCD7586FEDD3119C0B00A0C9E4605F6F99ED at uklozms02.Teleglobe.CA>, "Ha llgren, Michael" writes: > > >Robert Kiessling wrote: > >> Nurani Nimpuno writes: >> > We would >> > therefore like to request the community to consider making it >> > mandatory for NEW installations to use domain based web-hosting, with >> > the exception of a set of agreed applications needing IP based >> > web-hosting (eg. SSL). >> >> So you would require servers like www.siemens.de to share an IP >> address with www.my-completely-unimportant-personal-site.de if there >> is no SSL or similar use? I think if you interpret this loosely as "Only one IP per interface" you will get the right idea. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk at FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD coreteam member | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. From clive at demon.net Wed May 10 11:39:08 2000 From: clive at demon.net (Clive D.W. Feather) Date: Wed, 10 May 2000 10:39:08 +0100 Subject: IP assignment for virtual webhosting In-Reply-To: <200005100129.VAA23828@spare.de.teleglobe.net>; from briand@teleglobe.net on Tue, May 09, 2000 at 09:29:45PM -0400 References: <6BF1C330AF53D311BE5D00508B09081001E3FE3B@PLANET01> <200005100129.VAA23828@spare.de.teleglobe.net> Message-ID: <20000510103908.K72047@demon.net> Brian Dickson said: > I think there are a number of issues concerning the definition of "NEW". > > Perhaps the only one to which most will agree is a square-one startup, > with no pre-existing hardware, software, or address space. Sorry, but I *don't* agree. Put cynically (and I know that the people proposing this aren't being cynical), this says: The people in the business right now are introducing a rule that places new entrants at a disadvantage. This will prevent them from competing on an equal footing. This is called a "cartel" in most places, and is illegal in several countries (including the UK and, I believe, all of the EU) unless it can be specifically justified. -- Clive D.W. Feather | Work: | Tel: +44 20 8371 1138 Internet Expert | Home: | Fax: +44 20 8371 1037 Demon Internet | WWW: http://www.davros.org | DFax: +44 20 8371 4037 Thus plc | | Mobile: +44 7973 377646 From clive at demon.net Wed May 10 11:40:28 2000 From: clive at demon.net (Clive D.W. Feather) Date: Wed, 10 May 2000 10:40:28 +0100 Subject: IP assignment for virtual webhosting In-Reply-To: ; from joao@ripe.net on Tue, May 09, 2000 at 10:53:09AM +0200 References: <009E9CB4.A21D38B8.9@cc.univie.ac.at> Message-ID: <20000510104028.L72047@demon.net> Joao Luis Silva Damas said: > We'll be happy to assemble the list and will propose all applications > we see ourselves as needing IP based hosting, but it would be > impossible for us to come up with a complete list of all the > imaginative applications of technology that ISPs are using. It's not an issue of technologies. As Brian Dickson pointed out, the "slashdot effect" is enough to make you want to move a site to a new host or IP address. -- Clive D.W. Feather | Work: | Tel: +44 20 8371 1138 Internet Expert | Home: | Fax: +44 20 8371 1037 Demon Internet | WWW: http://www.davros.org | DFax: +44 20 8371 4037 Thus plc | | Mobile: +44 7973 377646 From niek at knoware.nl Wed May 10 12:28:49 2000 From: niek at knoware.nl (Niek Rijnbout) Date: Wed, 10 May 2000 12:28:49 +0200 Subject: IP assignment for virtual webhosting (how to move namebased virt host) In-Reply-To: <200005100129.VAA23828@spare.de.teleglobe.net> References: Message-ID: <4.3.1.2.20000510121355.00d9bdb0@bobo.knoware.nl> At 21:29 9-05-00 -0400, Brian Dickson wrote: >The issue with non-IP based virtuals, is the difficulty and complexity, and >more important, time-sensitivity involved, in *relocating* between servers. >DNS-based virtuals are impacted by caching resolvers, which are ubiquitous. >Lowering timers has risk too, especially if load spikes occur, eg for longer >than timer values. > >If everyone is aware of the "slashdot" effect, then you can appreciate wanting >to move such a "slashdotted" site to a separate server (or server farm), in >a timeframe less than the DNS normal-TTL-value times. Not doing so would have >a serious negative impact on 100 little-domains, who may have SLA's with you. > >Replace "slashdot" with "IPO" or "new software release", or any of a dozen >other usage-spike events; each customer has their own traffic trigger, and >the ISP won't necessarily have adequate warning. Three days is way too long; >even three hours can result in bad press, angry customers, and cancelled >contracts. How to move a hostname based virtual webserver: I have to webservers A and B (10.0.0.1) and (10.0.0.2) and want to move site www.cust.com from A to B The config on A is at the moment ServerName www.cust.com DocumentRoot ... Now move the data to B, change the nameserver entry for www.cust.com to point to 10.0.0.2 and create a www2.cust.com also to point to 10.0.0.2. The config file on B becomes ServerName www.cust.com ServerAlias www2.cust.com DocumentRoot ... and the config on A becomes ServerName www.cust.com Redirect / http://www2.cust.com/ So all request still coming in into A are redirected to the new name www2 (no DNS cache) and the DNS servers that are up to date go directly to B. Wait some time and remove the www2 entry from the DNS and clean up the config files. Hope this helps somewhat Regards Niek Rijnbout Knoware. From woeber at cc.univie.ac.at Wed May 10 16:26:20 2000 From: woeber at cc.univie.ac.at (Wilfried Woeber, UniVie/ACOnet) Date: Wed, 10 May 2000 16:26:20 +0200 Subject: We cannot really stop anyone from becoming a LIR... -- RIPE Message-ID: <009E9E24.F070C3F6.18@cc.univie.ac.at> >Certainly the only place I ever see an LIR name used is in the 'X-NCC-RegID' >header, which I'm sure isn't seen by customers. How are similar LIR names >going to confuse the general public when they never see them? ...and even there, the individual tags are qualified by the ISO3166 country codes. I'd be *really* surprised if the NCC would have allocated the same name twice for the *same* country?! -WW _________________________________:_____________________________________ Wilfried Woeber : e-mail: Woeber at CC.UniVie.ac.at UniVie Computer Center - ACOnet : Tel: +43 1 4277 - 140 33 Universitaetsstrasse 7 : Fax: +43 1 4277 - 9 140 A-1010 Vienna, Austria, Europe : RIPE-DB: WW144, PGP keyID 0xF0ACB369 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ From eamonn at ripe.net Wed May 10 16:37:43 2000 From: eamonn at ripe.net (Eamonn McGuinness) Date: Wed, 10 May 2000 16:37:43 +0200 Subject: We cannot really stop anyone from becoming a LIR... -- RIPE In-Reply-To: Your message of Wed, 10 May 2000 16:26:20 +0200. <009E9E24.F070C3F6.18@cc.univie.ac.at> References: <009E9E24.F070C3F6.18@cc.univie.ac.at> Message-ID: <200005101437.QAA06831@birch.ripe.net> You're right Wilfried. One company is called "Services", the other "Online". Both registries are not in the same country but are in the UK. We also used very different registry-IDs as well. Cheers, Eamonn McGuinness RIPE NCC Hostmaster In message <009E9E24.F070C3F6.18 at cc.univie.ac.at>you write: >>Certainly the only place I ever see an LIR name used is in the 'X-NCC-RegID' >>header, which I'm sure isn't seen by customers. How are similar LIR names >>going to confuse the general public when they never see them? > > ...and even there, the individual tags are qualified by the ISO3166 > country codes. I'd be *really* surprised if the NCC would have allocated > the same name twice for the *same* country?! > > -WW > _________________________________:_____________________________________ > Wilfried Woeber : e-mail: Woeber at CC.UniVie.ac.at > UniVie Computer Center - ACOnet : Tel: +43 1 4277 - 140 33 > Universitaetsstrasse 7 : Fax: +43 1 4277 - 9 140 > A-1010 Vienna, Austria, Europe : RIPE-DB: WW144, PGP keyID 0xF0ACB369 > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ From javier at bitmailer.com Thu May 11 13:41:58 2000 From: javier at bitmailer.com (Javier Llopis) Date: Thu, 11 May 2000 11:41:58 +0000 Subject: IP assignment for virtual webhosting Message-ID: <20000511094212.39DF125C8AB@proxy.bitmailer.com> On Mon, 08 May 2000 18:26:19 +0200, Nurani Nimpuno wrote: >We do however wish to raise our concern regarding what we see as an >inefficient usage of addresses in our limited address pool. We would >therefore like to request the community to consider making it >mandatory for NEW installations to use domain based web-hosting, with >the exception of a set of agreed applications needing IP based >web-hosting (eg. SSL). There is one issue I'd like to bring up that we constantly run into and was never brought up in this debate, which somehow amazes me. Our typical situation is that when a customer orders a domain named web site it comes along with an FTP server and a POP-3 server so they can have their own email addresses. While we could host all customer pages on the same host with the same IP using HTTP 1.1, in order to do the same with the POP servers we need to distinguish the POP server by its single IP address, since there is no HTTP 1.1 equivalent for POP. So we end up with 99% of our domains having a different IP address each, in which case who _cares_ if we also use that IP address for the web server? Are we really wasting IP address space? Hasn't anybody run into this situation? If so, how are you dealing with it? We would ask that, unless a viable solution to this problem is found, name based hosting should be encouraged but not required. BTW we are adopting HTTP 1.1 name based virtual hosts anyway in case the FTP and POP issues could be solved. Regards Javier Llopis BitMailer, S.L. javier at bitmailer.com Juan Bravo 51, Dup. 1-Izq Tel: +34 91 402 1551 28006 Madrid Fax: +34 91 402 4115 SPAIN From henning.brauer at bsmail.de Thu May 11 11:50:33 2000 From: henning.brauer at bsmail.de (henning.brauer at bsmail.de) Date: Thu, 11 May 2000 11:50:33 +0200 Subject: IP assignment for virtual webhosting Message-ID: Hi Javier, why do you use an own POP3 and FTP for each customer/domain???? We are using the same POP3/FTP/SMTP for all customers. Of course you need individual Logins, you can solve this by naming them CUSTOMERNR-1, -2 and so on or by using DOMAINNAME-1, DOMAINNAME--2 and so on. For ftp, we do a CHROOT in the customers Webroot (=his home-dir). So where is the problem?? Henning Brauer Hostmaster BSWS ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 PLEASE USE EMAIL WHERE POSSIBLE "Javier Llopis" To: "lir-wg at ripe.net" Subject: Re: IP assignment for virtual webhosting Sent by: owner-lir-wg@ ripe.net 11.05.00 13:41 Please respond to "Javier Llopis" On Mon, 08 May 2000 18:26:19 +0200, Nurani Nimpuno wrote: >We do however wish to raise our concern regarding what we see as an >inefficient usage of addresses in our limited address pool. We would >therefore like to request the community to consider making it >mandatory for NEW installations to use domain based web-hosting, with >the exception of a set of agreed applications needing IP based >web-hosting (eg. SSL). There is one issue I'd like to bring up that we constantly run into and was never brought up in this debate, which somehow amazes me. Our typical situation is that when a customer orders a domain named web site it comes along with an FTP server and a POP-3 server so they can have their own email addresses. While we could host all customer pages on the same host with the same IP using HTTP 1.1, in order to do the same with the POP servers we need to distinguish the POP server by its single IP address, since there is no HTTP 1.1 equivalent for POP. So we end up with 99% of our domains having a different IP address each, in which case who _cares_ if we also use that IP address for the web server? Are we really wasting IP address space? Hasn't anybody run into this situation? If so, how are you dealing with it? We would ask that, unless a viable solution to this problem is found, name based hosting should be encouraged but not required. BTW we are adopting HTTP 1.1 name based virtual hosts anyway in case the FTP and POP issues could be solved. Regards Javier Llopis BitMailer, S.L. javier at bitmailer.com Juan Bravo 51, Dup. 1-Izq Tel: +34 91 402 1551 28006 Madrid Fax: +34 91 402 4115 SPAIN From javier at bitmailer.com Thu May 11 14:22:08 2000 From: javier at bitmailer.com (Javier Llopis) Date: Thu, 11 May 2000 12:22:08 +0000 Subject: IP assignment for virtual webhosting Message-ID: <20000511102220.5610F25C8FE@proxy.bitmailer.com> On Thu, 11 May 2000 11:55:25 +0200 (CEST), Nils Jeppe wrote: >Use one big POP3 server for all of your customers, and either use >"standardized" pop-login names (ie, customercorp01, customercorp02, >othercorp 01, othercorp02, etc) and entries in the alias file to point to >them. It even makes sense to point mail.customerdomain.com as a cname to >your mail.provider.whatever so that if there's a change you don't have to >reconfigure all MUA's. Mmmm... A customer having two addresses in two different files/databases: That works but might be a nightmare to update. Javier Llopis BitMailer, S.L. javier at bitmailer.com Juan Bravo 51, Dup. 1-Izq Tel: +34 91 402 1551 28006 Madrid Fax: +34 91 402 4115 SPAIN From noc at entire-systems.com Thu May 11 12:27:58 2000 From: noc at entire-systems.com (Daniel Roesen) Date: Thu, 11 May 2000 12:27:58 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: ; from nils@work.de on Thu, May 11, 2000 at 11:55:25AM +0200 References: <20000511094212.39DF125C8AB@proxy.bitmailer.com> Message-ID: <20000511122758.A31725@entire-systems.com> On Thu, May 11, 2000 at 11:55:25AM +0200, Nils Jeppe wrote: > It even makes sense to point mail.customerdomain.com as a cname to > your mail.provider.whatever so that if there's a change you don't > have to reconfigure all MUA's. No. RFC 1912: "Common DNS Operational and Configuration Errors" 2.4 CNAME records [...] Don't use CNAMEs in combination with RRs which point to other names like MX, CNAME, PTR and NS. (PTR is an exception if you want to implement classless in-addr delegation.) For example, this is strongly discouraged: podunk.xx. IN MX mailhost mailhost IN CNAME mary mary IN A 1.2.3.4 [RFC 1034] in section 3.6.2 says this should not be done, and [RFC 974] explicitly states that MX records shall not point to an alias defined by a CNAME. This results in unnecessary indirection in accessing the data, and DNS resolvers and servers need to work more to get the answer. If you really want to do this, you can accomplish the same thing by using a preprocessor such as m4 on your host files. > As for ftp, it's easily solved by creating different home directories > for different logins Not, if you want to give your customers anonymous FTP space. Best regards, Daniel -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 232 bytes Desc: not available URL: From nils at work.de Thu May 11 12:25:55 2000 From: nils at work.de (Nils Jeppe) Date: Thu, 11 May 2000 12:25:55 +0200 (CEST) Subject: IP assignment for virtual webhosting In-Reply-To: <20000511102220.5610F25C8FE@proxy.bitmailer.com> Message-ID: On Thu, 11 May 2000, Javier Llopis wrote: > Mmmm... A customer having two addresses in two different files/databases: > That works but might be a nightmare to update. Nah, not really... it's still better than having to ervice/update/maintain dozens of independent mail servers. Nils - ----------------------------------------------------------------- - n at work Internet Informationssysteme GmbH Tel +49 40 23880900 Spaldingstrasse 160d Fax +49 40 23880929 20097 Hamburg, Germany http://www.work.de/ From nils at work.de Thu May 11 11:55:25 2000 From: nils at work.de (Nils Jeppe) Date: Thu, 11 May 2000 11:55:25 +0200 (CEST) Subject: IP assignment for virtual webhosting In-Reply-To: <20000511094212.39DF125C8AB@proxy.bitmailer.com> Message-ID: On Thu, 11 May 2000, Javier Llopis wrote: > there is no HTTP 1.1 equivalent for POP. So we end up with 99% of our > domains having a different IP address each, in which case who _cares_ > if we also use that IP address for the web server? > > Are we really wasting IP address space? > > Hasn't anybody run into this situation? If so, how are you dealing > with it? Use one big POP3 server for all of your customers, and either use "standardized" pop-login names (ie, customercorp01, customercorp02, othercorp 01, othercorp02, etc) and entries in the alias file to point to them. It even makes sense to point mail.customerdomain.com as a cname to your mail.provider.whatever so that if there's a change you don't have to reconfigure all MUA's. As for ftp, it's easily solved by creating different home directories for different logins (don't forget to chroot them). So your customers will not get their own hostname in the login prompt, no big deal.... ;-) Nils - ----------------------------------------------------------------- - n at work Internet Informationssysteme GmbH Tel +49 40 23880900 Spaldingstrasse 160d Fax +49 40 23880929 20097 Hamburg, Germany http://www.work.de/ From noc at entire-systems.com Thu May 11 12:31:36 2000 From: noc at entire-systems.com (Daniel Roesen) Date: Thu, 11 May 2000 12:31:36 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: <20000511094212.39DF125C8AB@proxy.bitmailer.com>; from javier@bitmailer.com on Thu, May 11, 2000 at 11:41:58AM +0000 References: <20000511094212.39DF125C8AB@proxy.bitmailer.com> Message-ID: <20000511123136.B31725@entire-systems.com> On Thu, May 11, 2000 at 11:41:58AM +0000, Javier Llopis wrote: > There is one issue I'd like to bring up that we constantly run into > and was never brought up in this debate, which somehow amazes me. And another one, not raised (enough): Accounting Best regards, Daniel -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 232 bytes Desc: not available URL: From henning.brauer at bsmail.de Thu May 11 12:40:54 2000 From: henning.brauer at bsmail.de (henning.brauer at bsmail.de) Date: Thu, 11 May 2000 12:40:54 +0200 Subject: IP assignment for virtual webhosting Message-ID: I have written about accounting, but no one replied! once again: Hi all, there is one big problem i see: If you charge your customers for traffic, you must count the traffic, and the easiest and most reliable way to do this is on the firewall/router. Thats the way we do and therefore we need one IP per customer. Most customers have more than one site and for that we are using HTTP 1.1 (NameVirtualHosts). As long as there is no Package for apache to count the traffic !reliable! (the traffic counted via access logs is << real traffic ) I do not see a way to put all customers to a couple of IPs. Greetings from Germany Henning Brauer Hostmaster BSWS ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 PLEASE USE EMAIL WHERE POSSIBLE Daniel Roesen stems.com> cc: Sent by: Subject: Re: IP assignment for virtual webhosting owner-lir-wg at r ipe.net 11.05.00 12:31 On Thu, May 11, 2000 at 11:41:58AM +0000, Javier Llopis wrote: > There is one issue I'd like to bring up that we constantly run into > and was never brought up in this debate, which somehow amazes me. And another one, not raised (enough): Accounting Best regards, Daniel -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: C.DTF Type: application/octet-stream Size: 242 bytes Desc: not available URL: From mark at pine.nl Thu May 11 13:22:38 2000 From: mark at pine.nl (Mark Lastdrager) Date: Thu, 11 May 2000 13:22:38 +0200 (MET DST) Subject: IP assignment for virtual webhosting In-Reply-To: Message-ID: At Thu, 11 May 2000, owner-lir-wg at ripe.net wrote: >there is one big problem i see: If you charge your customers for traffic, >you must count the traffic, and the easiest and most reliable way to do >this is on the firewall/router. Thats the way we do and therefore we need >one IP per customer. Most customers have more than one site and for that we >are using HTTP 1.1 (NameVirtualHosts). >As long as there is no Package for apache to count the traffic !reliable! >(the traffic counted via access logs is << real traffic ) I do not see a >way to put all customers to a couple of IPs. Same over here. We have built our own database-driven system over the last years which links IP adresses to customers, measures their traffic and even has the ability to traffic-shape groups of IP adresses to a certain bandwith, like a Packeteer box can do. Input comes from the border router which has IP accounting. Ofcourse when customers want more than one hostname linked to the same virthost (but with different content) we use a name-based virthost. Because of this system it is very hard to implement name-based virtual hosting, all accounting is done on one IP adress then and we have to rethink our accounting scheme of virthosts (which takes time, costs money etc. etc.) Mark Lastdrager Pine Internet -- email: mark at lastdrager.nl :: ML1400-RIPE :: tel. +31-70-3111010 http://www.pine.nl :: RIPE RegID nl.pine :: fax. +31-70-3111011 PGP key ID 92BB81D1 :: Dutch security news @ http://security.nl Today's excuse: Dew on the telephone lines. From dave.wilson at heanet.ie Thu May 11 12:51:35 2000 From: dave.wilson at heanet.ie (Dave Wilson) Date: Thu, 11 May 2000 11:51:35 +0100 Subject: IP assignment for virtual webhosting In-Reply-To: ; from henning.brauer@bsmail.de on Thu, May 11, 2000 at 12:40:54PM +0200 References: Message-ID: <20000511115135.R26100@urda.heanet.ie> Henning, > (the traffic counted via access logs is << real traffic ) Can you expand on this? Is the difference between counted and real traffic random, or a roughly equal (or proportional) overhead for each server? Is the difference so significant that one cannot take it into account as part of the charging structure? Regards, Dave -- dave.wilson at heanet.ie --------------------------------------- +353-1-662-3412 It is one thing to pray; it is another to pray to entities who might not only be listening, but who will search you out on the road and beat you across the head with sticks if you say something that offends them. -- Neil Gaiman ------------------ For crypto key send a blank message to davew+pgp at heanet.ie From michael.hallgren at teleglobe.com Thu May 11 13:18:04 2000 From: michael.hallgren at teleglobe.com (Hallgren, Michael) Date: Thu, 11 May 2000 12:18:04 +0100 Subject: IP assignment for virtual webhosting Message-ID: <04BCD7586FEDD3119C0B00A0C9E4605F6F9B72@uklozms02.Teleglobe.CA> >I have written about accounting, but no one replied! > In a previous life I became familiar with NFR (http://www.nfr.net/), and used it for some accounting tasks (based on other keys than IP addresses). I don't believe I was the only one, but can't remember where I found other people discussing it - if you're interested, I'll dig through my old mail. Then IPv6 :) mh >once again: Hi all, there is one big problem i see: If you charge your customers for traffic, you must count the traffic, and the easiest and most reliable way to do this is on the firewall/router. Thats the way we do and therefore we need one IP per customer. Most customers have more than one site and for that we are using HTTP 1.1 (NameVirtualHosts). As long as there is no Package for apache to count the traffic !reliable! (the traffic counted via access logs is << real traffic ) I do not see a way to put all customers to a couple of IPs. Greetings from Germany Henning Brauer Hostmaster BSWS ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 PLEASE USE EMAIL WHERE POSSIBLE Daniel Roesen stems.com> cc: Sent by: Subject: Re: IP assignment for virtual webhosting owner-lir-wg at r ipe.net 11.05.00 12:31 On Thu, May 11, 2000 at 11:41:58AM +0000, Javier Llopis wrote: > There is one issue I'd like to bring up that we constantly run into > and was never brought up in this debate, which somehow amazes me. And another one, not raised (enough): Accounting Best regards, Daniel -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 From dave.wilson at heanet.ie Thu May 11 12:40:18 2000 From: dave.wilson at heanet.ie (Dave Wilson) Date: Thu, 11 May 2000 11:40:18 +0100 Subject: IP assignment for virtual webhosting In-Reply-To: <20000511122758.A31725@entire-systems.com>; from noc@entire-systems.com on Thu, May 11, 2000 at 12:27:58PM +0200 References: <20000511094212.39DF125C8AB@proxy.bitmailer.com> <20000511122758.A31725@entire-systems.com> Message-ID: <20000511114018.P26100@urda.heanet.ie> > > It even makes sense to point mail.customerdomain.com as a cname to > > your mail.provider.whatever so that if there's a change you don't > > have to reconfigure all MUA's. > > No. > > RFC 1912: "Common DNS Operational and Configuration Errors" I believe the intention here is that primary MX for the domain is mail.provider.com (A record, easily changed at the DNS), and the SMTP server setting for customers' MUAs is mail.customerdomain.com (valid use of CNAME, easily changed at the DNS instead of manual reconfiguration). Regards, Dave -- dave.wilson at heanet.ie --------------------------------------- +353-1-662-3412 It is one thing to pray; it is another to pray to entities who might not only be listening, but who will search you out on the road and beat you across the head with sticks if you say something that offends them. -- Neil Gaiman ------------------ For crypto key send a blank message to davew+pgp at heanet.ie From noc at entire-systems.com Thu May 11 13:30:29 2000 From: noc at entire-systems.com (Entire Systems NOC) Date: Thu, 11 May 2000 13:30:29 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: ; from mark@pine.nl on Thu, May 11, 2000 at 01:22:38PM +0200 References: Message-ID: <20000511133029.A360@entire-systems.com> On Thu, May 11, 2000 at 01:22:38PM +0200, Mark Lastdrager wrote: > Because of this system it is very hard to implement name-based virtual > hosting, all accounting is done on one IP adress then and we have to > rethink our accounting scheme of virthosts (which takes time, costs money > etc. etc.) Exactly. You basically have to either a) analyze webserver logs (very inaccurate, CPU-intensive) b) write your own accounting software basing on sniffing, analyzing packet payloads for HTTP 1.1 header information and tracing TCP streams. This can also lead to a restructuring of you networks to be able to sniff. AFAIK Strato (hosting >500.000 .de domains) does something like b) 'cause they can't afford spending CPU time on their hosting server to do any accounting... (but this is hearsay). Best regards, Daniel Roesen Entire Systems NOC -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 232 bytes Desc: not available URL: From noc at entire-systems.com Thu May 11 13:35:53 2000 From: noc at entire-systems.com (Entire Systems NOC) Date: Thu, 11 May 2000 13:35:53 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: <20000511114018.P26100@urda.heanet.ie>; from dave.wilson@heanet.ie on Thu, May 11, 2000 at 11:40:18AM +0100 References: <20000511094212.39DF125C8AB@proxy.bitmailer.com> <20000511122758.A31725@entire-systems.com> <20000511114018.P26100@urda.heanet.ie> Message-ID: <20000511133553.A457@entire-systems.com> On Thu, May 11, 2000 at 11:40:18AM +0100, Dave Wilson wrote: > I believe the intention here is that primary MX for the domain is > mail.provider.com (A record, easily changed at the DNS), and the > SMTP server setting for customers' MUAs is mail.customerdomain.com Ah yes. I've overseen "M*U*A". Sure, that's fine. Best regards, Daniel Roesen Entire Systems NOC -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 232 bytes Desc: not available URL: From noc at entire-systems.com Thu May 11 13:50:32 2000 From: noc at entire-systems.com (Entire Systems NOC) Date: Thu, 11 May 2000 13:50:32 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: ; from henning.brauer@bsmail.de on Thu, May 11, 2000 at 01:35:29PM +0200 References: Message-ID: <20000511135032.B457@entire-systems.com> On Thu, May 11, 2000 at 01:35:29PM +0200, henning.brauer at bsmail.de wrote: > And if you think of Bandwith management - i don't see any way > to do this on an per name-basis, only per ip (and service) is > possible as far as I know. For Apache, mod_throttle (http://www.snert.com/Software/Throttle/) should do the trick. No idea wether it's realiable/CPU-hogging. And this doesn't cover any other, non-HTTP services. Best regards, Daniel Roesen Entire Systems NOC -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 232 bytes Desc: not available URL: From henning.brauer at bsmail.de Thu May 11 13:46:17 2000 From: henning.brauer at bsmail.de (henning.brauer at bsmail.de) Date: Thu, 11 May 2000 13:46:17 +0200 Subject: IP assignment for virtual webhosting Message-ID: it's not random. For simple static websites it's about 10-20%. What seems to be not counted is the data sent from the browser to the server, so if you have bigger interactive systems where lots of data ist entered or uploaded by our visitors, the difference is really high. I could do some checks to give you exact data; we did so long ago and decided that there is no good way to do an not-ip-based Accounting. We now have our own Database driven accounting & billing system which would be really complicated if i had to read out all the webserver logs. It would be very cool to have a traffic counter as part of apache (=>module) which stores the traffic on a by-named-host basis. Then Life would be much easier and we would need less IP-Space... (anyone here who has the time and knowledge to write such a module? =;-)))))) ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 PLEASE USE EMAIL WHERE POSSIBLE Dave Wilson cc: Sent by: Subject: Re: IP assignment for virtual webhosting owner-lir-wg@ ripe.net 11.05.00 12:51 Henning, > (the traffic counted via access logs is << real traffic ) Can you expand on this? Is the difference between counted and real traffic random, or a roughly equal (or proportional) overhead for each server? Is the difference so significant that one cannot take it into account as part of the charging structure? Regards, Dave -- dave.wilson at heanet.ie --------------------------------------- +353-1-662-3412 It is one thing to pray; it is another to pray to entities who might not only be listening, but who will search you out on the road and beat you across the head with sticks if you say something that offends them. -- Neil Gaiman ------------------ For crypto key send a blank message to davew+pgp at heanet.ie From mark at pine.nl Thu May 11 14:13:00 2000 From: mark at pine.nl (Mark Lastdrager) Date: Thu, 11 May 2000 14:13:00 +0200 (MET DST) Subject: IP assignment for virtual webhosting In-Reply-To: Message-ID: At Thu, 11 May 2000, Pavel Golubev wrote: >It's really sad, but your apache-around discussion is not interesting for >90% mailing list readers... Are you sure that I (and all my collegue) want >read THIS? We're ready to keep our eyes open for ANY RIPE team messages, >even 1st April jokes, but NOT your life-around discussion about >bananas prices in Honduras. The discussion is about the problems with HTTP 1.1 hosting (which RIPE suggests because it safes address space). I support the ideas of RIPE on this matter, but want to know if anybody has come around the problems with the accounting. If you are not interested in this topic, please filter it. Furthermore I suggest that you obey the netiquette rules (e.g. learn how to quote). Ofcourse the banana prices in Honduras have my interest too, but I'm on a separate mailing list for that. >To RIPE team: Sirs, please, open about-life at ripe.net mailing list >discussion for this kind of writer :) 'haha' Mark Lastdrager Pine Internet -- email: mark at lastdrager.nl :: ML1400-RIPE :: tel. +31-70-3111010 http://www.pine.nl :: RIPE RegID nl.pine :: fax. +31-70-3111011 PGP key ID 92BB81D1 :: Dutch security news @ http://security.nl Today's excuse: boss forgot system password From henning.brauer at bsmail.de Thu May 11 13:54:34 2000 From: henning.brauer at bsmail.de (henning.brauer at bsmail.de) Date: Thu, 11 May 2000 13:54:34 +0200 Subject: IP assignment for virtual webhosting Message-ID: Strato is not a good example... for a), you pointed out the important thing: It's inaccurate. If it would only be CPU-Power, you could use an own machine fopr that. forb), the main problem is to write such a software. Who of us has the time to do so? It would be very cool to have such a piece of software, but who can develop it? Or is there something that is capable to do so? Second problem, as you noticed, there could be a need to change your betwork layout. Our backbone is switched, and not every switch (in the Server-Racks we don't use the expensive Ciscos, there are simple Switches without Monitor-Port). So you would have to place the sniffer a) on the firewalls or the routers -> not a good idea - think of CPU, Diskspace, .. or b) but not always possible put in in the transfernet between firewalls and borderrouters. b) could be a olution for us if there were some kind of software usable for that. Greetings Henning Brauer Hostmaster BSWS ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 PLEASE USE EMAIL WHERE POSSIBLE Entire Systems NOC To: "lir-wg at ripe.net" Subject: Re: IP assignment for virtual webhosting Sent by: owner-lir-wg at r ipe.net 11.05.00 13:30 On Thu, May 11, 2000 at 01:22:38PM +0200, Mark Lastdrager wrote: > Because of this system it is very hard to implement name-based virtual > hosting, all accounting is done on one IP adress then and we have to > rethink our accounting scheme of virthosts (which takes time, costs money > etc. etc.) Exactly. You basically have to either a) analyze webserver logs (very inaccurate, CPU-intensive) b) write your own accounting software basing on sniffing, analyzing packet payloads for HTTP 1.1 header information and tracing TCP streams. This can also lead to a restructuring of you networks to be able to sniff. AFAIK Strato (hosting >500.000 .de domains) does something like b) 'cause they can't afford spending CPU time on their hosting server to do any accounting... (but this is hearsay). Best regards, Daniel Roesen Entire Systems NOC -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: C.DTF Type: application/octet-stream Size: 242 bytes Desc: not available URL: From henning.brauer at bsmail.de Thu May 11 13:57:58 2000 From: henning.brauer at bsmail.de (henning.brauer at bsmail.de) Date: Thu, 11 May 2000 13:57:58 +0200 Subject: IP assignment for virtual webhosting Message-ID: For me traffic management means to ensure that some really important sites always have enough bandwith and therefore first limiting bandwith for mail, ftp and so on.Second point is to make sure that even if there is a DoS-attack are something like this that uses all available bandwith that i can make sure to have a usabel SSH-Connection to the hosts (or in general: make sure that there is IN EVERY SITUATION enough bandwith for the netmanagement). ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 PLEASE USE EMAIL WHERE POSSIBLE Mark Lastdrager To: henning.brauer at bsmail.de Subject: Re: IP assignment for virtual webhosting 11.05.00 13:41 At Thu, 11 May 2000, henning.brauer at bsmail.de wrote: >Yes, that exactly the way we do it. If I imagine the power needed to >implement this its nearly impossible to reimpleent for ONLY name-based >Virtual Hosts. And if you think of Bandwith management - i don't see any >way to do this on an per name-basis, only per ip (and service) is possible >as far as I know. To use Bandwidth mangement on named vhosts-basis, you >would have to read out the FULL ip-header and service-specific headers on >the router...i think this is not possible without enormous CPU-power and >would give high delays on the router. There is something called 'mod_throttle' (if a remember well) for Apache, but I don't think customers are very interested in paying a flat fee for a virthost instead of being fast all the time and paying per MB. Mark Lastdrager Pine Internet -- email: mark at lastdrager.nl :: ML1400-RIPE :: tel. +31-70-3111010 http://www.pine.nl :: RIPE RegID nl.pine :: fax. +31-70-3111011 PGP key ID 92BB81D1 :: Dutch security news @ http://security.nl Today's excuse: quatnum decoherence From colinj at uk.psi.com Thu May 11 13:47:30 2000 From: colinj at uk.psi.com (Colin Johnston) Date: Thu, 11 May 2000 12:47:30 +0100 (BST) Subject: IP assignment for virtual webhosting In-Reply-To: Message-ID: > as far as I know. To use Bandwidth mangement on named vhosts-basis, you > would have to read out the FULL ip-header and service-specific headers on > the router...i think this is not possible without enormous CPU-power and > would give high delays on the router. > ------------------------------------------------ This is possible using F5Labs equipment however I dont belive this has been fully implemented on large web farms as yet but maybe soon :) Colin Johnston SA PSINET UK From mark at pine.nl Thu May 11 13:44:34 2000 From: mark at pine.nl (Mark Lastdrager) Date: Thu, 11 May 2000 13:44:34 +0200 (MET DST) Subject: IP assignment for virtual webhosting In-Reply-To: <20000511115135.R26100@urda.heanet.ie> Message-ID: At Thu, 11 May 2000, owner-lir-wg at ripe.net wrote: >Can you expand on this? Is the difference between counted and real >traffic random, or a roughly equal (or proportional) overhead for >each server? Is the difference so significant that one cannot take >it into account as part of the charging structure? I did some research on this topic a while ago. The traffic Apache measures (based on filesize) can be 10 to 50% smaller than the traffic the router measures. Ofcourse this depends on the filesize, packetsize etc. Mark Lastdrager Pine Internet -- email: mark at lastdrager.nl :: ML1400-RIPE :: tel. +31-70-3111010 http://www.pine.nl :: RIPE RegID nl.pine :: fax. +31-70-3111011 PGP key ID 92BB81D1 :: Dutch security news @ http://security.nl Today's excuse: the router thinks its a printer. From pg at gu.net Thu May 11 14:04:26 2000 From: pg at gu.net (Pavel Golubev) Date: Thu, 11 May 2000 15:04:26 +0300 (EEST) Subject: IP assignment for virtual webhosting In-Reply-To: Message-ID: On Thu, 11 May 2000, Mark Lastdrager wrote: Dear Sirs, It's really sad, but your apache-around discussion is not interesting for 90% mailing list readers... Are you sure that I (and all my collegue) want read THIS? We're ready to keep our eyes open for ANY RIPE team messages, even 1st April jokes, but NOT your life-around discussion about bananas prices in Honduras. To RIPE team: Sirs, please, open about-life at ripe.net mailing list discussion for this kind of writer :) > Date: Thu, 11 May 2000 13:44:34 +0200 (MET DST) > From: Mark Lastdrager > To: Dave Wilson > Cc: lir-wg at ripe.net > Subject: Re: IP assignment for virtual webhosting > > At Thu, 11 May 2000, owner-lir-wg at ripe.net wrote: > > >Can you expand on this? Is the difference between counted and real > >traffic random, or a roughly equal (or proportional) overhead for > >each server? Is the difference so significant that one cannot take > >it into account as part of the charging structure? > > I did some research on this topic a while ago. The traffic Apache measures > (based on filesize) can be 10 to 50% smaller than the traffic the router > measures. Ofcourse this depends on the filesize, packetsize etc. > > Mark Lastdrager > Pine Internet > > -- > email: mark at lastdrager.nl :: ML1400-RIPE :: tel. +31-70-3111010 > http://www.pine.nl :: RIPE RegID nl.pine :: fax. +31-70-3111011 > PGP key ID 92BB81D1 :: Dutch security news @ http://security.nl > Today's excuse: the router thinks its a printer. > > --- Pavel Golubev PG810-RIPE Global Ukraine Ltd. From mark at pine.nl Thu May 11 14:07:09 2000 From: mark at pine.nl (Mark Lastdrager) Date: Thu, 11 May 2000 14:07:09 +0200 (MET DST) Subject: IP assignment for virtual webhosting In-Reply-To: Message-ID: At Thu, 11 May 2000, owner-lir-wg at ripe.net wrote: >for a), you pointed out the important thing: It's inaccurate. If it would >only be CPU-Power, you could use an own machine fopr that. >forb), the main problem is to write such a software. Who of us has the time >to do so? It would be very cool to have such a piece of software, but who >can develop it? Or is there something that is capable to do so? >Second problem, as you noticed, there could be a need to change your >betwork layout. Our backbone is switched, and not every switch (in the >Server-Racks we don't use the expensive Ciscos, there are simple Switches >without Monitor-Port). So you would have to place the sniffer a) on the >firewalls or the routers -> not a good idea - think of CPU, Diskspace, .. >or b) but not always possible put in in the transfernet between firewalls >and borderrouters. b) could be a olution for us if there were some kind of >software usable for that. We have a machine over here that shares a HUB with the border router. That machine (which has no IP address ofcourse ;-)) thus sees all traffic and is our IDS. It could do accounting too if neccessary, but then there indeed has to be software which reads out HTTP 1.1 traffic. Mark Lastdrager Pine Internet -- email: mark at lastdrager.nl :: ML1400-RIPE :: tel. +31-70-3111010 http://www.pine.nl :: RIPE RegID nl.pine :: fax. +31-70-3111011 PGP key ID 92BB81D1 :: Dutch security news @ http://security.nl Today's excuse: Standing room only on the bus. From Havard.Eidnes at runit.sintef.no Thu May 11 14:00:31 2000 From: Havard.Eidnes at runit.sintef.no (Havard.Eidnes at runit.sintef.no) Date: Thu, 11 May 2000 14:00:31 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: Your message of "Thu, 11 May 2000 11:40:18 +0100" <20000511114018.P26100@urda.heanet.ie> References: <20000511114018.P26100@urda.heanet.ie> Message-ID: <20000511140031O.he@runit.sintef.no> > > RFC 1912: "Common DNS Operational and Configuration Errors" > > I believe the intention here is that primary MX for the domain is > mail.provider.com (A record, easily changed at the DNS), and the > SMTP server setting for customers' MUAs is mail.customerdomain.com > (valid use of CNAME, easily changed at the DNS instead of manual > reconfiguration). That's a valid use of CNAME records in conjunction with mail handling, but about the only one. You still shouldn't combine MX and CNAME, which is what RFC1912 talks about in the previously quoted section. However, the text from RFC1912 is still valid (the example lacks the MX priority field, though ;-), but doesn't point out what the reason for the RFC 974 restriction is. So if you'll excuse me for flying off on a DNS tangent... The mail server presents itself with one name in its greeting message. Hopefully this name is a fully qualified domain name (probably mail.provider.com in the above example). This is the name you should be using as the target in an MX record when you want to use the mail service of that host. If you adhere to this rule, your mail server will correctly handle being a "secondary" mail handler for a domain (modulo any configured relaying restrictions), and recognize "itself" in a list of MX targets, and prune off itself and any lower- prioritized entries. If you don't, well..., you're doing something which is similar to walking around pointing a loaded and unsecured gun in the direction of your own foot. - H?vard From henning.brauer at bsmail.de Thu May 11 13:35:29 2000 From: henning.brauer at bsmail.de (henning.brauer at bsmail.de) Date: Thu, 11 May 2000 13:35:29 +0200 Subject: IP assignment for virtual webhosting Message-ID: Yes, that exactly the way we do it. If I imagine the power needed to implement this its nearly impossible to reimpleent for ONLY name-based Virtual Hosts. And if you think of Bandwith management - i don't see any way to do this on an per name-basis, only per ip (and service) is possible as far as I know. To use Bandwidth mangement on named vhosts-basis, you would have to read out the FULL ip-header and service-specific headers on the router...i think this is not possible without enormous CPU-power and would give high delays on the router. ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 PLEASE USE EMAIL WHERE POSSIBLE Mark Lastdrager To: henning.brauer at bsmail.de , "lir-wg at ripe.net" > Subject: Re: IP assignment for virtual webhosting 11.05.00 13:22 At Thu, 11 May 2000, owner-lir-wg at ripe.net wrote: >there is one big problem i see: If you charge your customers for traffic, >you must count the traffic, and the easiest and most reliable way to do >this is on the firewall/router. Thats the way we do and therefore we need >one IP per customer. Most customers have more than one site and for that we >are using HTTP 1.1 (NameVirtualHosts). >As long as there is no Package for apache to count the traffic !reliable! >(the traffic counted via access logs is << real traffic ) I do not see a >way to put all customers to a couple of IPs. Same over here. We have built our own database-driven system over the last years which links IP adresses to customers, measures their traffic and even has the ability to traffic-shape groups of IP adresses to a certain bandwith, like a Packeteer box can do. Input comes from the border router which has IP accounting. Ofcourse when customers want more than one hostname linked to the same virthost (but with different content) we use a name-based virthost. Because of this system it is very hard to implement name-based virtual hosting, all accounting is done on one IP adress then and we have to rethink our accounting scheme of virthosts (which takes time, costs money etc. etc.) Mark Lastdrager Pine Internet -- email: mark at lastdrager.nl :: ML1400-RIPE :: tel. +31-70-3111010 http://www.pine.nl :: RIPE RegID nl.pine :: fax. +31-70-3111011 PGP key ID 92BB81D1 :: Dutch security news @ http://security.nl Today's excuse: Dew on the telephone lines. From mally at theplanet.net Thu May 11 14:23:30 2000 From: mally at theplanet.net (Mally Mclane) Date: Thu, 11 May 2000 13:23:30 +0100 Subject: IP assignment for virtual webhosting Message-ID: <6BF1C330AF53D311BE5D00508B09081001A024F2@PLANET01> We use FSLabs equipment across most of our major client regards, mally mclane =>hostmaster/core systems =>for the Planet Online, Leeds, UK > -----Original Message----- > From: Colin Johnston [mailto:colinj at uk.psi.com] > Sent: Thursday, May 11, 2000 12:48 PM > To: henning.brauer at bsmail.de > Cc: lir-wg at ripe.net > Subject: Re: IP assignment for virtual webhosting > > > > as far as I know. To use Bandwidth mangement on named > vhosts-basis, you > > would have to read out the FULL ip-header and > service-specific headers on > > the router...i think this is not possible without enormous > CPU-power and > > would give high delays on the router. > > ------------------------------------------------ > > This is possible using F5Labs equipment however I dont belive this has > been fully implemented on large web farms as yet but maybe soon :) > > Colin Johnston > SA > PSINET UK > > From noc at entire-systems.com Thu May 11 14:49:11 2000 From: noc at entire-systems.com (Daniel Roesen) Date: Thu, 11 May 2000 14:49:11 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: ; from pg@gu.net on Thu, May 11, 2000 at 03:04:26PM +0300 References: Message-ID: <20000511144911.A1049@entire-systems.com> On Thu, May 11, 2000 at 03:04:26PM +0300, Pavel Golubev wrote: > It's really sad, but your apache-around discussion is not > interesting for 90% mailing list readers... From: Nurani Nimpuno To: lir-wg at ripe.net Subject: IP assignment for virtual webhosting [...] We wish to add this to the agenda at the upcoming RIPE meeting for further discussion and welcome any input the community may have on this matter. > Are you sure that I (and all my collegue) want read THIS? man procmail Daniel Roesen Entire Systems NOC -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 232 bytes Desc: not available URL: From mally at theplanet.net Thu May 11 14:31:31 2000 From: mally at theplanet.net (Mally Mclane) Date: Thu, 11 May 2000 13:31:31 +0100 Subject: IP assignment for virtual webhosting Message-ID: <6BF1C330AF53D311BE5D00508B09081001A024F5@PLANET01> {and the bit I accidentally deleted) ...our major clients, (freeserve, lloydstsb, jungle.com etc) but not on vhosts. M > -----Original Message----- > From: Mally Mclane [mailto:mally at theplanet.net] > Sent: Thursday, May 11, 2000 1:24 PM > To: 'Colin Johnston'; 'lir-wg at ripe.net' > Subject: RE: IP assignment for virtual webhosting > > > We use FSLabs equipment across most of our major client > > regards, > > mally mclane > =>hostmaster/core systems > =>for the Planet Online, Leeds, UK > > > > -----Original Message----- > > From: Colin Johnston [mailto:colinj at uk.psi.com] > > Sent: Thursday, May 11, 2000 12:48 PM > > To: henning.brauer at bsmail.de > > Cc: lir-wg at ripe.net > > Subject: Re: IP assignment for virtual webhosting > > > > > > > as far as I know. To use Bandwidth mangement on named > > vhosts-basis, you > > > would have to read out the FULL ip-header and > > service-specific headers on > > > the router...i think this is not possible without enormous > > CPU-power and > > > would give high delays on the router. > > > ------------------------------------------------ > > > > This is possible using F5Labs equipment however I dont > belive this has > > been fully implemented on large web farms as yet but maybe soon :) > > > > Colin Johnston > > SA > > PSINET UK > > > > > From herbert at hostit.be Thu May 11 16:09:45 2000 From: herbert at hostit.be (Herbert Baerten) Date: Thu, 11 May 2000 16:09:45 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: Message-ID: <02ee01bfbb52$8f225ed0$01040a0a@herbert.hostit.be> > why do you use an own POP3 and FTP for each customer/domain???? > We are using the same POP3/FTP/SMTP for all customers. Of course you need > individual Logins, you can solve this by naming them CUSTOMERNR-1, -2 and > so on or by using DOMAINNAME-1, DOMAINNAME--2 and so on. For ftp, we do a > CHROOT in the customers Webroot (=his home-dir). > So where is the problem?? We are also using one singel pop3 and smtp server for all customers, and also use the CHROOT solution for FTP accounts on un*x servers, however other FTP servers on other platforms (don't want to start an OS war here, but some of us do use IIS on NT for FTP :)) do not always support this mechanism. Has anyone solved this on NT? Perhaps by using different FTP server software? And I agree with Daniel that accounting is a very important issue here. Security is another one, firewalls being configured based on source and destination IP addresses. kind regards, Herbert -- Herbert Baerten HB5351 HostIT Network Manager NCC9166-RIPE From noc at entire-systems.com Thu May 11 16:27:27 2000 From: noc at entire-systems.com (Daniel Roesen) Date: Thu, 11 May 2000 16:27:27 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: <02ee01bfbb52$8f225ed0$01040a0a@herbert.hostit.be>; from herbert@hostit.be on Thu, May 11, 2000 at 04:09:45PM +0200 References: <02ee01bfbb52$8f225ed0$01040a0a@herbert.hostit.be> Message-ID: <20000511162727.A1592@entire-systems.com> On Thu, May 11, 2000 at 04:09:45PM +0200, Herbert Baerten wrote: > And I agree with Daniel that accounting is a very important issue here. > Security is another one, firewalls being configured based on source and > destination IP addresses. And another (although minor) point: you loose flexibilty. Situation: you want to move a bunch of domains hosted on server A to another server B because of system load or whatever. With name-based vhosting you have to do the usual DNS changes and wait for DNS convergency (DNS cache timeouts) to start the move. With IP-based vhosting you can act in minutes. Move config over to server B, take down IP on server A, activate IP on server B and you're set. This means in the name-based vhosting case i have about a week (usual RR TTL) before being able to react on the slashdot effect in contrary to IP-based vhosting where I'm able to react in minutes. I already saw several well known sites suffering from exactly THIS problem. Especially for companies like us with very limited upstream bandwidth. Best regards, Daniel Roesen Entire Systems NOC -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 232 bytes Desc: not available URL: From noc at entire-systems.com Thu May 11 16:33:41 2000 From: noc at entire-systems.com (Daniel Roesen) Date: Thu, 11 May 2000 16:33:41 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: <20000511162727.A1592@entire-systems.com>; from noc@entire-systems.com on Thu, May 11, 2000 at 04:27:27PM +0200 References: <02ee01bfbb52$8f225ed0$01040a0a@herbert.hostit.be> <20000511162727.A1592@entire-systems.com> Message-ID: <20000511163341.A2070@entire-systems.com> On Thu, May 11, 2000 at 04:27:27PM +0200, Daniel Roesen wrote: > This means in the name-based vhosting case i have about a week (usual > RR TTL) before being able to react on the slashdot effect in contrary Sorry, i meant "1 to several days". Daniel Roesen Entire Systems NOC -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 232 bytes Desc: not available URL: From noc at entire-systems.com Thu May 11 16:51:26 2000 From: noc at entire-systems.com (Daniel Roesen) Date: Thu, 11 May 2000 16:51:26 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: ; from henning.brauer@bsmail.de on Thu, May 11, 2000 at 04:37:08PM +0200 References: Message-ID: <20000511165126.A2311@entire-systems.com> On Thu, May 11, 2000 at 04:37:08PM +0200, henning.brauer at bsmail.de wrote: > There was a solution for this on this list a few days ago... Sure, but you're screwed if you you auto-generate your zonefiles and VHost-configs from databases and you want to move many VHosts... Best regards, Daniel Roesen Entire Systems NOC -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 232 bytes Desc: not available URL: From henning.brauer at bsmail.de Thu May 11 17:01:36 2000 From: henning.brauer at bsmail.de (henning.brauer at bsmail.de) Date: Thu, 11 May 2000 17:01:36 +0200 Subject: IP assignment for virtual webhosting Message-ID: thats our situation too. doing some changes in the scripts should do the things. ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 PLEASE USE EMAIL WHERE POSSIBLE Daniel Roesen cc: Sent by: Subject: Re: IP assignment for virtual webhosting owner-lir-wg at r ipe.net 11.05.00 16:51 On Thu, May 11, 2000 at 04:37:08PM +0200, henning.brauer at bsmail.de wrote: > There was a solution for this on this list a few days ago... Sure, but you're screwed if you you auto-generate your zonefiles and VHost-configs from databases and you want to move many VHosts... Best regards, Daniel Roesen Entire Systems NOC -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: C.DTF Type: application/octet-stream Size: 242 bytes Desc: not available URL: From henning.brauer at bsmail.de Thu May 11 16:37:08 2000 From: henning.brauer at bsmail.de (henning.brauer at bsmail.de) Date: Thu, 11 May 2000 16:37:08 +0200 Subject: IP assignment for virtual webhosting Message-ID: There was a solution for this on this list a few days ago... How to move a hostname based virtual webserver: I have to webservers A and B (10.0.0.1) and (10.0.0.2) and want to move site www.cust.com from A to B The config on A is at the moment ServerName www.cust.com DocumentRoot ... Now move the data to B, change the nameserver entry for www.cust.com to point to 10.0.0.2 and create a www2.cust.com also to point to 10.0.0.2. The config file on B becomes ServerName www.cust.com ServerAlias www2.cust.com DocumentRoot ... and the config on A becomes ServerName www.cust.com Redirect / http://www2.cust.com/ So all request still coming in into A are redirected to the new name www2 (no DNS cache) and the DNS servers that are up to date go directly to B. Wait some time and remove the www2 entry from the DNS and clean up the config files. ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 PLEASE USE EMAIL WHERE POSSIBLE Daniel Roesen cc: Sent by: Subject: Re: IP assignment for virtual webhosting owner-lir-wg at r ipe.net 11.05.00 16:27 On Thu, May 11, 2000 at 04:09:45PM +0200, Herbert Baerten wrote: > And I agree with Daniel that accounting is a very important issue here. > Security is another one, firewalls being configured based on source and > destination IP addresses. And another (although minor) point: you loose flexibilty. Situation: you want to move a bunch of domains hosted on server A to another server B because of system load or whatever. With name-based vhosting you have to do the usual DNS changes and wait for DNS convergency (DNS cache timeouts) to start the move. With IP-based vhosting you can act in minutes. Move config over to server B, take down IP on server A, activate IP on server B and you're set. This means in the name-based vhosting case i have about a week (usual RR TTL) before being able to react on the slashdot effect in contrary to IP-based vhosting where I'm able to react in minutes. I already saw several well known sites suffering from exactly THIS problem. Especially for companies like us with very limited upstream bandwidth. Best regards, Daniel Roesen Entire Systems NOC -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: C.DTF Type: application/octet-stream Size: 242 bytes Desc: not available URL: From juhas at cabinet.net Thu May 11 17:02:33 2000 From: juhas at cabinet.net (Juha Suhonen) Date: Thu, 11 May 2000 18:02:33 +0300 (EEST) Subject: IP assignment for virtual webhosting In-Reply-To: <20000511165126.A2311@entire-systems.com> Message-ID: On Thu, 11 May 2000, Daniel Roesen wrote: > On Thu, May 11, 2000 at 04:37:08PM +0200, henning.brauer at bsmail.de wrote: > > There was a solution for this on this list a few days ago... > Sure, but you're screwed if you you auto-generate your zonefiles and > VHost-configs from databases and you want to move many VHosts... And this solution also breaks some web sites - you can't redirect POSTs (imagine a site that has forms linking to it from all over the web - you simply can't change them all). And besides, I've seen many sites that use full-blown URLs in their local
tags. And, after this kind of a change, you have to keep the www2.example.com "alive" forever. (and I'm not even talking about the customer's point of view - he certainly isn't pleased to find out that suddenly his website has mystically "moved" from www.example.com to www2.example.com, as said by the Address -field of his web browser) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Juha Suhonen THK-Net Cabinet Group, Systems Administrator Tel +358 9 6126 580 Fax +358 9 61265899 http://www.thk.net ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ From noc at entire-systems.com Thu May 11 17:24:44 2000 From: noc at entire-systems.com (Daniel Roesen) Date: Thu, 11 May 2000 17:24:44 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: <20000511165126.A2311@entire-systems.com>; from noc@entire-systems.com on Thu, May 11, 2000 at 04:51:26PM +0200 References: <20000511165126.A2311@entire-systems.com> Message-ID: <20000511172444.A2441@entire-systems.com> On Thu, May 11, 2000 at 04:51:26PM +0200, Daniel Roesen wrote: > Sure, but you're screwed if you you auto-generate your zonefiles > and VHost-configs from databases and you want to move many VHosts... And it bites you right were it hurts most, if your customers website depends in any way on absolute links to http://www.foo.bar/, so www.foo.bar still gets all request and has to send redirects. Worst case: customer uses all-absolute links and depends on "www.foo.bar" in CGIs parsing HTTP Host infos or has some htaccess tricks or whatever. I'm not an Apache expert at all so others may comment on this. I have a horror about moving xx sites to another host and seeing some percentage being broken. Daniel Roesen Entire Systems NOC -- Entire Systems Network Operations Center noc at entire-systems.com Entire Systems GmbH - Ferbachstrasse 12 - 56203 Hoehr-Grenzhausen, Germany InterNIC-Handle: ES1238-ORG RIPE-Handle: ESN10-RIPE Tel: +49 2624 9550-55 GnuPG/PGP Key-ID: 0xBF3C40C9 http://www.entire-systems.com/noc/noc-key.asc GnuPG/PGP Fingerprint: 1F3F B675 1A38 D87C EB3C 6090 C6B9 DF48 BF3C 40C9 -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 232 bytes Desc: not available URL: From henning.brauer at bsmail.de Thu May 11 17:09:34 2000 From: henning.brauer at bsmail.de (henning.brauer at bsmail.de) Date: Thu, 11 May 2000 17:09:34 +0200 Subject: IP assignment for virtual webhosting Message-ID: Mike, this was a bit misunderstandable as i wrote a few hours ago. traffic management is most important for different services than for different vservers. our "add an module..." requires a) to write such a module (i do not have the time and the knowledge to do so) and b) doesn't solve the problems pointed out here before (ip paket sizes for expample). if you could have a look to our server farm you would see that the "adding hardware" aproach is not our way. ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 PLEASE USE EMAIL WHERE POSSIBLE Miquel van Smoorenburg To: henning.brauer at bsmail.de ron.nl> Subject: Re: IP assignment for virtual webhosting 11.05.00 17:04 According to henning.brauer at bsmail.de: > Yes, that exactly the way we do it. If I imagine the power needed to > implement this its nearly impossible to reimpleent for ONLY name-based > Virtual Hosts. Add an apache log module that measures the actual bytes sent out (file size + HTTP headers) - accounting problem solved. > And if you think of Bandwith management - i don't see any > way to do this on an per name-basis, only per ip (and service) is possible > as far as I know. To use Bandwidth mangement on named vhosts-basis, you > would have to read out the FULL ip-header and service-specific headers on > the router...i think this is not possible without enormous CPU-power and > would give high delays on the router. Use the Apache module mod_bandwidth, or something similar. People tend to buy and install enormous amounts of hardware, solving problems by adding more boxes. While fixing the software is simpler, easier and faster. Mike. From miquels at cistron.nl Thu May 11 17:04:27 2000 From: miquels at cistron.nl (Miquel van Smoorenburg) Date: Thu, 11 May 2000 17:04:27 +0200 Subject: IP assignment for virtual webhosting In-Reply-To: ; from henning.brauer@bsmail.de on Thu, May 11, 2000 at 01:35:29PM +0200 References: Message-ID: <20000511170427.B32298@cistron.nl> According to henning.brauer at bsmail.de: > Yes, that exactly the way we do it. If I imagine the power needed to > implement this its nearly impossible to reimpleent for ONLY name-based > Virtual Hosts. Add an apache log module that measures the actual bytes sent out (file size + HTTP headers) - accounting problem solved. > And if you think of Bandwith management - i don't see any > way to do this on an per name-basis, only per ip (and service) is possible > as far as I know. To use Bandwidth mangement on named vhosts-basis, you > would have to read out the FULL ip-header and service-specific headers on > the router...i think this is not possible without enormous CPU-power and > would give high delays on the router. Use the Apache module mod_bandwidth, or something similar. People tend to buy and install enormous amounts of hardware, solving problems by adding more boxes. While fixing the software is simpler, easier and faster. Mike. From javier at bitmailer.com Thu May 11 14:45:50 2000 From: javier at bitmailer.com (Javier Llopis) Date: Thu, 11 May 2000 12:45:50 +0000 Subject: IP assignment for virtual webhosting Message-ID: <200005111756.TAA10117@birch.ripe.net> On Thu, 11 May 2000 12:25:55 +0200 (CEST), Nils Jeppe wrote: >> Mmmm... A customer having two addresses in two different files/databases: >> That works but might be a nightmare to update. > >Nah, not really... it's still better than having to ervice/update/maintain >dozens of independent mail servers. But we DO have only one big POP server, the only thing is that it looks the IP address the client is connecting to to know which domain she belongs to. The alias approach is workable, but I foresee update inconsistencies created over time which can be a pain in the neck. But even then I'd have go to hundreds of organizations and thousands of users to tell them to reconfigure their email programs in a way that they won't understand (believe me). And I would have to tell them that in order to use the handy web-mail gateway they have to type an awkard user name, instead of 'my address is someuser at somecompany.es' I can put up with the update problem since that is an internal matter of my organization, but I don't want this kind of things to reflect outside nor I want customers to change their simple way of doing things if not for the better (for them, not for us or RIPE). Javier Llopis BitMailer, S.L. javier at bitmailer.com Juan Bravo 51, Dup. 1-Izq Tel: +34 91 402 1551 28006 Madrid Fax: +34 91 402 4115 SPAIN From henning.brauer at bsmail.de Fri May 12 01:41:55 2000 From: henning.brauer at bsmail.de (henning.brauer at bsmail.de) Date: Fri, 12 May 2000 01:41:55 +0200 Subject: Summary: IP assignment for virtual webhosting Message-ID: -from the http side of view, it's ok to use name based vhosts -problem: accounting There is no software package to do an !reliable! accounting on a name-based basis, Accounting per IP works fine Possible solution is a sniffer. this requires an extra machine (and perhaps a second machine more for analyzing the collected data) and perhaps changes in the network layout. There is no software package to do this. -problem: bandwidth management Bandwidth management on an ip + service based environment is possible and not tooooo complicated. For the http part, there are some packages for apache, but they have not all features needed and you cannot manage badwith for other services than http. -problem: moving vhosts DNS is not fast enough, the possible solution with a redirect to a new www2-host doesnt work in all cases. So i think it is not possible to use only one ip per machine and doing only name based hosting. I think our way with one IP per customer (NOT per vhost) is an acceptable compromise. I won't be at the RIPE-Meeting. Could someone please point out a summary of our discussion? Greetings from Germany Henning Brauer Hostmaster BSWS ------------------------------------------------ BS Web Services Roedingsmarkt 14 20459 Hamburg Germany info at bsmail.de www.bsws.de fon: +49 40 3750357-0 fax: +49 40 3750357-5 From joao at ripe.net Fri May 12 14:04:44 2000 From: joao at ripe.net (Joao Luis Silva Damas) Date: Fri, 12 May 2000 14:04:44 +0200 Subject: Summary: IP assignment for virtual webhosting In-Reply-To: References: Message-ID: At 1:41 +0200 12/5/00, henning.brauer at bsmail.de wrote: >I won't be at the RIPE-Meeting. Could someone please point out a summary of >our discussion? Definitely, the RIPE NCC will, and thanks for the compilation e-mail with these points. Joao Damas Head of External Services RIPE NCC >Greetings from Germany > >Henning Brauer >Hostmaster BSWS From hph at online.no Mon May 15 00:23:50 2000 From: hph at online.no (Hans Petter Holen) Date: Mon, 15 May 2000 00:23:50 +0200 Subject: Draft Agenda v 2 Message-ID: <00c001bfbdf3$151280e0$e206e1c3@dont.no> Dear Working Group, I am terribly sorry I am so late with the preparations for next weeks RIPE meeting. This is mainly due to unforseen events on my side. Please find below a draft agenda for our upcoming working group meeting. The first version of this was presented to the plenary at the last RIPE meeting. I have removed item 6 the report from the address council item as this will be presented to the ICANN open meeting on friday. Be aware that I have not gone trough the mailinglist lately to se if there has been any requests there for items to be added. Pleas drop me a note if there are any other items that ought to be on the agenda. 1. Admin -scribe, participant list, charter, mailinglists 2. Agenda 3. Meet the RIPE NCC hostmasters 4. RIPE 35 - minutes - actions 5. Reports from the Registries - RIPE NCC - APNIC - ARIN - ICANN - Status of the Latin and AFRI NiCs 7. Presentations of candidates 8. Microallocations in the ARIN region. 9. IP addresses to GPRS infrastructure. 10. Interim report from the ICANN ad Hoc group 11. Minimum Allocation 12. Name-based webhosting Yours, Hans Petter Holen Local IR Working group Chair. ---- Hans Petter Holen, VP New Products, Scandinavia Online mailto:hph at a.sol.no Phone: +47 88 00 39 16 From hph at online.no Mon May 15 01:54:46 2000 From: hph at online.no (Hans Petter Holen) Date: Mon, 15 May 2000 01:54:46 +0200 Subject: Action list and Draft minutes v1 LIR-WG 35 Message-ID: <00f901bfbdff$c98af320$e206e1c3@dont.no> I have now updated my "Work in progress" pages available from the LIR-WG pages http://www.ripe.net/ripe/wg/lir/index.html The draft minutes as produced by Paul Tate from the RIPE NCC has been made available http://home.sol.no/~hph/RIPE/RIPE%2035%20minutes.htm My slides from the meeting http://home.sol.no/~hph/RIPE/LIR%20WG%2035_files/frame.htm and my report to the plenary http://home.sol.no/~hph/RIPE/LIR%20WG%2035%20plenary%20report-filer/frame.ht m Please submit comments and corrections to me or the list. The huge delay is entirely my fault. Many thanks to Paul for producing the minutes ! Action list: Action Owner Status Description 35.1 Chair Publish policy document 35.2 Chair Publish election procedure 35.3 WG Subscribe lir-wg 35.4 NCC PGP Key exchange procedure 35.5 NCC Implement PGP for hostmaster mail 35.6 NCC Make db checking tool available for self audit 35.7 Chair Task Force on addresses to GPRS infrastructure -hph From tt at byteaction.de Fri May 12 08:31:26 2000 From: tt at byteaction.de (Thomas Trede) Date: Fri, 12 May 2000 08:31:26 +0200 Subject: Summary: IP assignment for virtual webhosting References: Message-ID: <000b01bfbbdb$b3f87620$edf817c1@hrz.tudarmstadt.de> Hello, I am very sure, that either the WG-Chair or someone of the RIPE NCC will take care of that. Regards, Thomas Trede ----- Original Message ----- From: To: Sent: Friday, May 12, 2000 1:41 AM Subject: Summary: IP assignment for virtual webhosting > I won't be at the RIPE-Meeting. Could someone please point out a summary of > our discussion? From Werner.Erhard at cantrade.ch Thu May 11 12:35:34 2000 From: Werner.Erhard at cantrade.ch (Erhard Werner, ERW) Date: Thu, 11 May 2000 12:35:34 +0200 Subject: WG: Notification: Inbound Mail Failure Message-ID: > -----Urspr?ngliche Nachricht----- > Von: ExchangeAdmins > Gesendet am: Donnerstag, 11. Mai 2000 12:34 > An: ExchangeAdmins > Betreff: Notification: Inbound Mail Failure > > The following recipients did not receive the attached mail. A NDR was not > sent to the originator for the following recipients for one of the > following reasons: > > > * The Delivery Status Notification options did not request failure > notification, or requested no notification. > > * The message was of precedence bulk. > > > > NDR reasons are listed with each recipient, along with the notification > requested for that recipient, > or the precedence. > > felix.fischer at cantrade.ch > MSEXCH:IMS:CANTRADE:CAZH:CZEXCHANGE1 0 (000C05A6) Unknown Recipient > Precedence: bulk > > The message that caused this notification was: > > <> -------------- next part -------------- An embedded message was scrubbed... From: Daniel Roesen Subject: Re: IP assignment for virtual webhosting Date: Thu, 11 May 2000 12:27:58 +0200 Size: 2048 URL: From becha at ripe.net Mon May 15 11:52:25 2000 From: becha at ripe.net (Vesna Manojlovic) Date: Mon, 15 May 2000 11:52:25 +0200 Subject: Warning about the size of the NEXT message Message-ID: <200005150952.LAA28436@kantoor.ripe.net> Dear list members, Please be aware that the next message coming to the list has a Word document attachment. The attachment is rather large. It is not our practice to allow such huge messages to go to the list. We tried to find an alternative way of posting it; however, due to lack of time we did not succeed. It is important that this message reaches lir-wg before the RIPE meeting. The message is from Adrian Pauling, and the subject is "IP Adddressing & GPRS Mobile Technology deployment". Apologies once more for not being able to approve it to the list in any other way. Regards, Vesna Manojlovic RIPE NCC lists moderator From adrian.pauling at bt.com Wed May 10 17:15:52 2000 From: adrian.pauling at bt.com (adrian.pauling at bt.com) Date: Wed, 10 May 2000 16:15:52 +0100 Subject: IP Adddressing & GPRS Mobile Technology deployment Message-ID: <27EDC2145E42D211AD9600606DD5EC1D072888E4@mbrpb1nt02.mww.bt.com> Ladies & Gentlemen, At the RIPE35 meeting there was some debate regarding the requirements for IP addressing by the GPRS Mobile technology. A small working group was formed, and subsequently some progress has been made. The attachment provides more information, and will be of interest to those attending the LIR-WG at RIPE36. <> > Regards, > > Adrian F Pauling > :-)NEL2C Internet Protocol Manager > acd Information Systems Engineering Technical Architecture > AFP1-RIPE / AFP-ARIN / AFP25-InterNIC > * adrian.pauling at bt.com > * +44 19 2685 1992 / +44 78 0290 4877 > British Telecommunications plc > Registered Office 81 Newgate Street London EC1A 7AJ > Registered in England no 1800000 -------------- next part -------------- A non-text attachment was scrubbed... Name: Working_Party_Briefing_paper_GPRS_addressing.doc Type: application/msword Size: 461312 bytes Desc: not available URL: From henk at ripe.net Mon May 15 13:29:26 2000 From: henk at ripe.net (Henk Uijterwaal (RIPE-NCC)) Date: Mon, 15 May 2000 13:29:26 +0200 (CEST) Subject: Warning about the size of the NEXT message In-Reply-To: <200005151101.NAA00374@aegir.EU.net> Message-ID: On Mon, 15 May 2000, James Aldridge wrote: > Vesna Manojlovic wrote: > > Please be aware that the next message coming to the list has a Word > > document attachment. The attachment is rather large. > > You may want to take a look at Adri van Os's "antiword" program (for Unix, > RISC OS & BeOS) to convert Microsoft Word documents into somewhat more > reasonable formats (text and Postscript) (see http://www.winfield.demon.nl/ > for details). In this case it reduced the 600+ kbyte encoded attachment > (461,312 byte Word document) to an 8,040 byte text file: > 461312 May 15 13:48 Working_Party_Briefing_paper_GPRS_addressing.doc > 8040 May 15 13:48 Working_Party_Briefing_paper_GPRS_addressing.txt Warning: I tried that as well, then wondered what was in the other 450000 bytes, so I opened the document in Word. If you open the document in Word, you'll see a M$ Word icon indicating a second word document that is included in the first one. Clicking on the icon will open a second document with more text. Antiword doesn't recognize this, so the 8040 bytes is only part of the text. Henk ------------------------------------------------------------------------------ Henk Uijterwaal Email: henk.uijterwaal at ripe.net RIPE Network Coordination Centre WWW: http://www.ripe.net/home/henk Singel 258 Phone: +31.20.535-4414, Fax -4445 1016 AB Amsterdam Home: +31.20.4195305 The Netherlands Mobile: +31.6.55861746 ------------------------------------------------------------------------------ A man can take a train and never reach his destination. (Kerouac, well before RFC2780). From jhma at EU.net Mon May 15 13:01:13 2000 From: jhma at EU.net (James Aldridge) Date: Mon, 15 May 2000 13:01:13 +0200 Subject: Warning about the size of the NEXT message In-Reply-To: Your message of "Mon, 15 May 2000 11:52:25 +0200." <200005150952.LAA28436@kantoor.ripe.net> Message-ID: <200005151101.NAA00374@aegir.EU.net> Vesna Manojlovic wrote: > Dear list members, > > Please be aware that the next message coming to the list has a Word > document attachment. The attachment is rather large. > > It is not our practice to allow such huge messages to go to the list. We tried > to find an alternative way of posting it; however, due to lack of time we did > not succeed. It is important that this message reaches lir-wg before the RIPE > meeting. > > The message is from Adrian Pauling, and the subject is "IP Adddressing & GPRS > Mobile Technology deployment". > > Apologies once more for not being able to approve it to the list in any other > way. You may want to take a look at Adri van Os's "antiword" program (for Unix, RISC OS & BeOS) to convert Microsoft Word documents into somewhat more reasonable formats (text and Postscript) (see http://www.winfield.demon.nl/ for details). In this case it reduced the 600+ kbyte encoded attachment (461,312 byte Word document) to an 8,040 byte text file: 461312 May 15 13:48 Working_Party_Briefing_paper_GPRS_addressing.doc 8040 May 15 13:48 Working_Party_Briefing_paper_GPRS_addressing.txt I could really do without the extra, unnecessary ~600 kbytes in my mailbox... James -- James Aldridge Singel 540, 1017 AZ Amsterdam, NL Senior Network Engineer (R&D) Tel: +31 20 530 5327 KPNQwest GSM: +31 653 708 707 From john at ripe.net Mon May 15 13:30:34 2000 From: john at ripe.net (John Crain) Date: Mon, 15 May 2000 13:30:34 +0200 Subject: ANNOUNCEMENT: Beta 0.1 of asused-public released In-Reply-To: Message from Maldwyn Morris of "Fri, 12 May 2000 16:15:43 +0200." <200005121415.QAA12761@birch.ripe.net> Message-ID: <200005151130.NAA01358@birch.ripe.net> Dear Colleagues, We, the RIPE NCC, are pleased to announced the release of the beta version of asused-public. This is a tool to summarise address space that is registered in the RIPE database. For a given list of allocations the software prints various information concerning the allocations and the assignments they contain. the software "asused-public" is based on one the NCC's internal tools. It has been revised for public release in a form that does not make use of any customer confidential information. The software which uses several perl modules, these are included, is stored in the following file: ftp://ftp.ripe.net/tools/asused-public.0.1.tar.gz We are keen to receive your feedback on the beta release. Please address your comments and/or suggestions to our Software Manager, Maldwyn Morris Yours, John LeRoy Crain Head of Internal Services RIPE NCC From ash at ash.de Mon May 15 13:52:44 2000 From: ash at ash.de (Hauke Johannknecht) Date: Mon, 15 May 2000 13:52:44 +0200 (MEST) Subject: URLFIX: Beta 0.1 of asused-public released In-Reply-To: <200005151130.NAA01358@birch.ripe.net> Message-ID: On Mon, 15 May 2000, John Crain wrote: > The software which uses several perl modules, these are included, is > stored in the following file: the correct URL is: ftp://ftp.ripe.net/tools/asused-public-0.1b.tar.gz Gruss, Hauke -- Hauke Johannknecht Berlin / Germany HJ422-RIPE Use PGP ! -> lynx -dump http://www.ash.de/ash.asc | pgp -kaf From hph at online.no Mon May 15 13:22:11 2000 From: hph at online.no (Hans Petter Holen) Date: Mon, 15 May 2000 13:22:11 +0200 Subject: IP Adddressing & GPRS Mobile Technology deployment References: <27EDC2145E42D211AD9600606DD5EC1D072888E4@mbrpb1nt02.mww.bt.com> Message-ID: <003b01bfbe61$16546cc0$e306e1c3@dont.no> I have converted the document to HTML for thoose of you who cannot read MS attachments: http://home.sol.no/home/hph/RIPE/Working_Party_Briefing_paper_GPRS_addressin g.htm -hph -hph ----- Original Message ----- From: To: Sent: Wednesday, May 10, 2000 5:15 PM Subject: IP Adddressing & GPRS Mobile Technology deployment > Ladies & Gentlemen, > > At the RIPE35 meeting there was some debate regarding the requirements for > IP addressing by the GPRS Mobile technology. A small working group was > formed, and subsequently some progress has been made. The attachment > provides more information, and will be of interest to those attending the > LIR-WG at RIPE36. > <> > > > Regards, > > > > Adrian F Pauling > > :-)NEL2C Internet Protocol Manager > > acd Information Systems Engineering Technical Architecture > > AFP1-RIPE / AFP-ARIN / AFP25-InterNIC > > * adrian.pauling at bt.com > > * +44 19 2685 1992 / +44 78 0290 4877 > > British Telecommunications plc > > Registered Office 81 Newgate Street London EC1A 7AJ > > Registered in England no 1800000 > From jhma at EU.net Mon May 15 16:11:37 2000 From: jhma at EU.net (James Aldridge) Date: Mon, 15 May 2000 16:11:37 +0200 Subject: IP Adddressing & GPRS Mobile Technology deployment In-Reply-To: Your message of "Mon, 15 May 2000 13:22:11 +0200." <003b01bfbe61$16546cc0$e306e1c3@dont.no> Message-ID: <200005151411.QAA01107@aegir.EU.net> "Hans Petter Holen" wrote: > I have converted the document to HTML for thoose of you who cannot read MS > attachments: > > http://home.sol.no/home/hph/RIPE/Working_Party_Briefing_paper_GPRS_addressin > g.htm But that too is missing the attached documents. :-( I've put as complete a version as I can up on http://www.mcvax.org/~jhma/lir/gprs/ Regards, James -- James Aldridge Singel 540, 1017 AZ Amsterdam, NL Senior Network Engineer (R&D) Tel: +31 20 530 5327 KPNQwest GSM: +31 653 708 707 > -hph > > -hph > ----- Original Message ----- > From: > To: > Sent: Wednesday, May 10, 2000 5:15 PM > Subject: IP Adddressing & GPRS Mobile Technology deployment > > > > Ladies & Gentlemen, > > > > At the RIPE35 meeting there was some debate regarding the requirements for > > IP addressing by the GPRS Mobile technology. A small working group was > > formed, and subsequently some progress has been made. The attachment > > provides more information, and will be of interest to those attending the > > LIR-WG at RIPE36. > > <> > > > > > Regards, > > > > > > Adrian F Pauling > > > :-)NEL2C Internet Protocol Manager > > > acd Information Systems Engineering Technical Architecture > > > AFP1-RIPE / AFP-ARIN / AFP25-InterNIC > > > * adrian.pauling at bt.com > > > * +44 19 2685 1992 / +44 78 0290 4877 > > > British Telecommunications plc > > > Registered Office 81 Newgate Street London EC1A 7AJ > > > Registered in England no 1800000 > > From netmaster at space.net Mon May 15 12:34:57 2000 From: netmaster at space.net (Gert Doering, Netmaster) Date: Mon, 15 May 2000 12:34:57 +0200 Subject: Warning about the size of the NEXT message In-Reply-To: <200005150952.LAA28436@kantoor.ripe.net>; from becha@ripe.net on Mon, May 15, 2000 at 11:52:25AM +0200 References: <200005150952.LAA28436@kantoor.ripe.net> Message-ID: <20000515123457.G2836@Space.Net> Hi, On Mon, May 15, 2000 at 11:52:25AM +0200, Vesna Manojlovic wrote: > It is not our practice to allow such huge messages to go to the list. We tried > to find an alternative way of posting it; however, due to lack of time we did > not succeed. It is important that this message reaches lir-wg before the RIPE > meeting. A sensible way would have been to put it up on the RIPE web server, and just send the URL to the list. Besides this, sending winword is impolite anyway. I can't read it (no windows machine here) without having to ask a collegue from a different department to print it for me. If you want to send formatted document, politeness would ask for PDF. regards, Gert Doering -- NetMaster -- SpaceNet GmbH Mail: netmaster at Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299 From hank at att.net.il Mon May 15 14:46:03 2000 From: hank at att.net.il (Hank Nussbacher) Date: Mon, 15 May 2000 14:46:03 +0200 Subject: IP Adddressing & GPRS Mobile Technology deployment In-Reply-To: <27EDC2145E42D211AD9600606DD5EC1D072888E4@mbrpb1nt02.mww.bt .com> Message-ID: <3.0.5.32.20000515144603.007bdbc0@max.ibm.net.il> At 16:15 10/05/00 +0100, you wrote: Can you explain why it was sent on May 10 and only hit the list on May 15? -Hank >Ladies & Gentlemen, > >At the RIPE35 meeting there was some debate regarding the requirements for >IP addressing by the GPRS Mobile technology. A small working group was >formed, and subsequently some progress has been made. The attachment >provides more information, and will be of interest to those attending the >LIR-WG at RIPE36. > <> > >> Regards, >> >> Adrian F Pauling >> :-)NEL2C Internet Protocol Manager >> acd Information Systems Engineering Technical Architecture >> AFP1-RIPE / AFP-ARIN / AFP25-InterNIC >> * adrian.pauling at bt.com >> * +44 19 2685 1992 / +44 78 0290 4877 > From rgebhart at cybernet-ag.net Mon May 15 16:50:05 2000 From: rgebhart at cybernet-ag.net (Ralf Gebhart) Date: Mon, 15 May 2000 16:50:05 +0200 Subject: Warning about the size of the NEXT message In-Reply-To: <20000515123457.G2836@Space.Net>; from Gert Doering, Netmaster on Mon, May 15, 2000 at 12:34:57PM +0200 References: <200005150952.LAA28436@kantoor.ripe.net> <20000515123457.G2836@Space.Net> Message-ID: <20000515165005.N4793@cybernet-ag.net> On Mon, May 15, 2000 at 12:34:57PM +0200, Gert Doering, Netmaster wrote: > Hi, > > On Mon, May 15, 2000 at 11:52:25AM +0200, Vesna Manojlovic wrote: > > It is not our practice to allow such huge messages to go to the list. We tried > > to find an alternative way of posting it; however, due to lack of time we did > > not succeed. It is important that this message reaches lir-wg before the RIPE > > meeting. > > A sensible way would have been to put it up on the RIPE web server, and > just send the URL to the list. > > Besides this, sending winword is impolite anyway. I can't read it (no > windows machine here) without having to ask a collegue from a different > department to print it for me. If you want to send formatted document, > politeness would ask for PDF. Or HTML which is readable by even more systems than PDF. It's usually a 'one-mouse-click' action to create a proper HTML file within this M$ word app and you're done. -- Ralf Gebhart Network Planning & Realization Cybernet AG Munich From becha at ripe.net Mon May 15 17:34:24 2000 From: becha at ripe.net (Vesna Manojlovic) Date: Mon, 15 May 2000 17:34:24 +0200 Subject: IP Adddressing & GPRS Mobile Technology deployment In-Reply-To: Your message of Mon, 15 May 2000 14:46:03 +0200. <3.0.5.32.20000515144603.007bdbc0@max.ibm.net.il> References: <3.0.5.32.20000515144603.007bdbc0@max.ibm.net.il> Message-ID: <200005151534.RAA06868@kantoor.ripe.net> Hank Nussbacher writes: * At 16:15 10/05/00 +0100, you wrote: * * Can you explain why it was sent on May 10 and only hit the list on May 15? * Hello Henk, It was basically caused by a mixture of miscommunication and myself being a little preoccupied with RIPE Meeting preparations. My apologies for this. On May 10 Adrian sent the message, which didn't make it to the list - it was too large and bounced to moderator's mailbox. Instead of instantly approving it to the list, I asked Adrian to put the document on the web or ftp and to send a URL to the list - exactly for this reason: to avoid sending big attachments to the list. Unfortunately the answer from Adrian did not come to the moderator, myself, directly. As soon as we received his answer and realised we could best forward the large attachment the message was sent. Unfortunately this caused a delay which with hind sight may have been avoided. We sent the file with a warning, to inform people that the message contained a large MS format attachment. Becha RIPE NCC * -Hank * * >Ladies & Gentlemen, * > * >At the RIPE35 meeting there was some debate regarding the requirements for * >IP addressing by the GPRS Mobile technology. A small working group was * >formed, and subsequently some progress has been made. The attachment * >provides more information, and will be of interest to those attending the * >LIR-WG at RIPE36. * > <> * > * >> Regards, * >> * >> Adrian F Pauling * >> :-)NEL2C Internet Protocol Manager * >> acd Information Systems Engineering Technical Architecture * >> AFP1-RIPE / AFP-ARIN / AFP25-InterNIC * >> * adrian.pauling at bt.com * >> * +44 19 2685 1992 / +44 78 0290 4877 * > * * * From adrian.pauling at bt.com Mon May 15 16:32:18 2000 From: adrian.pauling at bt.com (adrian.pauling at bt.com) Date: Mon, 15 May 2000 15:32:18 +0100 Subject: IP Addressing & GPRS Mobile Technology deployment Message-ID: <27EDC2145E42D211AD9600606DD5EC1D03BD3195@mbrpb1nt02.mww.bt.com> Hank & LIR-WG, there was a delay due to the size of the attachment, which invoked the lists moderator, whilst the size and necessity of mailing such a file to the list were checked. The issue of IP Addressing & GPRS Mobile Technology deployment started at RIPE 35 and will be further discussed at RIPE 36. If time had permitted a more appropriate document format and probably web based URL for people to gain the information if they desired. My unreserved apologies for sending the file(s) in a proprietary document format - I was having trouble saving the embedded files as html. My thanks to Hans Petter Holen as LIR-WG Chair for providing a URL of the minutes, and James Aldridge of KPNQwest whose URL contains all the relevant details and document(s). Regards, Adrian F Pauling :-)NEL2C Internet Protocol Manager acd Information Systems Engineering Technical Architecture AFP1-RIPE / AFP-ARIN / AFP25-InterNIC * adrian.pauling at bt.com * +44 19 2685 1992 / +44 78 0290 4877 British Telecommunications plc Registered Office 81 Newgate Street London EC1A 7AJ Registered in England no 1800000 > -----Original Message----- > From: Hank Nussbacher [SMTP:hank at att.net.il] > Sent: 15 May 2000 13:46 > To: adrian.pauling at bt.com > Cc: lir-wg at ripe.net > Subject: Re: IP Adddressing & GPRS Mobile Technology deployment > > At 16:15 10/05/00 +0100, you wrote: > > Can you explain why it was sent on May 10 and only hit the list on May 15? > > -Hank > > >Ladies & Gentlemen, > > > >At the RIPE35 meeting there was some debate regarding the requirements > for > >IP addressing by the GPRS Mobile technology. A small working group was > >formed, and subsequently some progress has been made. The attachment > >provides more information, and will be of interest to those attending the > >LIR-WG at RIPE36. > > <> > > > >> Regards, > >> > >> Adrian F Pauling > >> :-)NEL2C Internet Protocol Manager > >> acd Information Systems Engineering Technical Architecture > >> AFP1-RIPE / AFP-ARIN / AFP25-InterNIC > >> * adrian.pauling at bt.com > >> * +44 19 2685 1992 / +44 78 0290 4877 > > From jorma.mellin at teliafi.net Tue May 16 08:04:18 2000 From: jorma.mellin at teliafi.net (Jorma Mellin) Date: Tue, 16 May 2000 09:04:18 +0300 Subject: Warning about the size of the NEXT message Message-ID: <200005160605.IAA13553@birch.ripe.net> >Or HTML which is readable by even more systems than PDF. >It's usually a 'one-mouse-click' action to create a proper HTML file >within this M$ word app and you're done. This M$ stuff shows how well these GSM people understand the way the Internet community works. And yet they are pleading addresses. I wonder why they are so keen to have publicly routable address space (from IPv4!) but not so worried about public AS numbers. To my ears this sounds that they haven't really though about this enough. Jorma Mellin ---------------------------------------------------------- jorma.mellin at teliafi.net Development Manager ; CCIE#4185 Telia Finland Inc, Network Services From Torsten.Blum at ecrc.de Tue May 16 06:46:03 2000 From: Torsten.Blum at ecrc.de (Torsten Blum) Date: Tue, 16 May 2000 06:46:03 +0200 (CEST) Subject: Warning about the size of the NEXT message In-Reply-To: <20000515123457.G2836@Space.Net> from "Gert Doering, Netmaster" at "May 15, 2000 12:34:57 pm" Message-ID: <200005160446.GAA15521@turon.ecrc.de> Gert Doering, Netmaster wrote: > A sensible way would have been to put it up on the RIPE web server, and > just send the URL to the list. > > Besides this, sending winword is impolite anyway. I can't read it (no > windows machine here) without having to ask a collegue from a different > department to print it for me. If you want to send formatted document, > politeness would ask for PDF. I definately agree. Regards Torsten Blum -- Torsten Blum, Network Engineer Tel: +49.89.92699.0 Cable & Wireless ECRC GmbH Fax: +49.89.92699.225 Arabellastr. 17 E-Mail: tblum at ecrc.de D-81925 Munich, Germany From stephenb at uk.uu.net Tue May 16 10:17:11 2000 From: stephenb at uk.uu.net (Stephen Burley) Date: Tue, 16 May 2000 08:17:11 +0000 Subject: Warning about the size of the NEXT message References: <200005160605.IAA13553@birch.ripe.net> Message-ID: <39210407.C28588A1@uk.uu.net> Jorma Mellin wrote: > >Or HTML which is readable by even more systems than PDF. > >It's usually a 'one-mouse-click' action to create a proper HTML file > >within this M$ word app and you're done. > > This M$ stuff shows how well these GSM people understand the way > the Internet community works. And yet they are pleading addresses. > I wonder why they are so keen to have publicly routable address space > (from IPv4!) but not so worried about public AS numbers. To my ears this > sounds that they haven't really though about this enough. > > Jorma Mellin On the contrary, it was thought through and discussed by all present at the working group. All issues where noted and as a hostmaster for a large ISP i tended to play devils advocate when it came to thorny questions like private address space or not and private root DNS or not. I am sure that a careful consideration of all the facts, by all who where not at the working party would see that all the relavent issues where looked into and reasonable answers were given. I agree for the most part the GSM people where not that Internet literate but once we put aside all preconceived ideas and got down to real engineering reasonings, the whole thing made sense. As for using public address space in a private enviroment without public AS numbers iit s very simple, private address space is limited to the ranges as detailed in the RFC, this network is going to be global and each network needs to be unique and using private address space would have caused logistical nightmares. For instance it is envisaged they whould run out of address space very quickly as 10.0.0.0/8 whould only be enough for Europe(as have UUNET / Worlcom internal networks), also if you have independant companies approaching the GPSM initiative from differant sides of the globe both using 10.0.0.0/8 and then want to connect to each other directly to enable roaming which company should renumber a multi million node GPSM network. But the use of private AS numbers is not nearly as difficult to administer or renumber to allow differant networks to talk to each other. Hope this helps. BTW i find it difficult to appreciate how the use of a piece of software to produce a technical report is a mesure of a persons RIPE compectance, i would rather say its a measure peoples unwillingness to accept another practical use of IP into the RIPE community. Lets work to help them into the community not allienate them from it. Regards, Stephen Burley UUNET EMEA Hostmaster > > > ---------------------------------------------------------- > jorma.mellin at teliafi.net > Development Manager ; CCIE#4185 > Telia Finland Inc, Network Services From jorma.mellin at teliafi.net Tue May 16 10:52:19 2000 From: jorma.mellin at teliafi.net (Jorma Mellin) Date: Tue, 16 May 2000 11:52:19 +0300 Subject: Warning about the size of the NEXT message Message-ID: <200005160853.KAA08661@birch.ripe.net> >As for using public address space in a private enviroment without public AS >numbers iit s very simple, private address space is limited to the ranges >as detailed in the RFC, this network is going to be global and each network >needs to be unique and using private address space would have caused >logistical nightmares. For instance it is envisaged they whould run out of >address space very quickly as 10.0.0.0/8 whould only be enough for >Europe(as have UUNET / Worlcom internal networks), also if you have >independant companies approaching the GPSM initiative from differant sides >of the globe both using 10.0.0.0/8 and then want to connect to each other >directly to enable roaming which company should renumber a multi million >node GPSM network. But the use of private AS numbers is not nearly as >difficult to administer or renumber to allow differant networks to talk to >each other. If the private address space mentioned in RFC is not enough they should go for IPv6. Because GPRS, UMTS and other stuff is brand new I see no problem to implement IPv6 in gateways (and terminals). Another issue is that roarming. Why roarming have to use unique IP numbers to work? You could use application level roarming solution as well and nobody cares if the source addr have gone trough a NAT box or not. This thing is going to get messy when we get first GPRS phones out. >BTW i find it difficult to appreciate how the use of a piece of software to >produce a technical report is a mesure of a persons RIPE compectance, i >would rather say its a measure peoples unwillingness to accept another >practical use of IP into the RIPE community. It's not about the software, it's about usability. I haven't found yet a Word application for my PDA, but I do have a browser :) Jorma From pgoncalves at mail.telepac.pt Wed May 17 17:26:06 2000 From: pgoncalves at mail.telepac.pt (Pedro =?iso-8859-1?Q?Gon=E7alves?=) Date: Wed, 17 May 2000 16:26:06 +0100 Subject: ANNOUNCEMENT: Beta 0.1 of asused-public released In-Reply-To: <200005151130.NAA01358@birch.ripe.net> References: <200005121415.QAA12761@birch.ripe.net> Message-ID: <4.1.20000517161800.009f3940@mail.telepac.pt> At 13:30 15-05-2000 +0200, John Crain wrote: > > >We are keen to receive your feedback on the beta release. > >Please address your comments and/or suggestions to our Software >Manager, Maldwyn Morris > Just to congratulate you for the release of this helpfull tool... No special "features" noticed... works fine for our alloc objects. ------------------------------------------------- Pedro Goncalves Telepac - Comunicacoes Interactivas DE 3 - Planeamento e Gestao Rede http://home.telepac.pt http://net.sapo.pt/ From ripe-dbm at ripe.net Wed May 17 18:30:58 2000 From: ripe-dbm at ripe.net (RIPE Database Administration) Date: Wed, 17 May 2000 18:30:58 +0200 Subject: Problems RIPE Whois Service. Message-ID: <200005171630.SAA25068@birch.ripe.net> Dear Colleagues, (Apologies if you receive this message more than once). Earlier this evening, the main RIPE Database server crashed. Your updates are not being processed at this time, but they are being queued; there is no need to send them again. Queries are being processed by a backup server; thus, the response time may be longer than usual. We are aware of the inconvience that this causes you and we are working to restore the service as quickly as possible. We shall keep you informed. If you have any questions, please contact . Regards, Ambrose Magee ------------- RIPE NCC From ripe-dbm at ripe.net Thu May 18 15:13:02 2000 From: ripe-dbm at ripe.net (RIPE Database Administration) Date: Thu, 18 May 2000 15:13:02 +0200 Subject: RIPE Database working normally again. Message-ID: <200005181313.PAA04720@birch.ripe.net> Dear Colleagues, Thank you for your patience. The RIPE Database is working normally again. Updates are being processed and queries are being answered by the primary server. If you have any questions, please contact . Daniele Arena ------------- RIPE NCC From pgl at petrel.net Thu May 18 16:51:22 2000 From: pgl at petrel.net (Pascal Gloor) Date: Thu, 18 May 2000 16:51:22 +0200 Subject: RIPE Database crashes ... Message-ID: <22CF090CD80ED41189CE00E018C23A00E45F@amonra.petrel.ch> Dear LIR's, What's up with the database ? Do you think it's normal that this machine crashes every 3-4 weeks ? Does RIPE need some more guys to manage that ? I think all LIR's whould be ok to pay RIPE a little bit more fee that RIPE may have a VERY HIGH BIGGER MTBF (Mean Time Between Failure). What do you think about that ? Or what else solution do you have guys (LIR's and RIPE)... to be continued .... Regards, Pascal PGL2-RIPE, AS6893, CH-PETREL Pascal Gloor Network Manager _________________________ SPAN(tm) / Petrel Communications SA A Cable & Wireless company T?l : + 41 22 304 47 47 Fax : + 41 22 304 47 99 E-mail : pgl at cw.span.ch Web : http://www.span.ch From mark at pine.nl Thu May 18 17:04:19 2000 From: mark at pine.nl (Mark Lastdrager) Date: Thu, 18 May 2000 17:04:19 +0200 (MET DST) Subject: RIPE Database crashes ... In-Reply-To: <22CF090CD80ED41189CE00E018C23A00E45F@amonra.petrel.ch> Message-ID: At Thu, 18 May 2000, owner-lir-wg at ripe.net wrote: >What do you think about that ? >Or what else solution do you have guys (LIR's and RIPE)... Well, maybe you can run a mirror database and give us the hostname? Mark Lastdrager Pine Internet -- email: mark at lastdrager.nl :: ML1400-RIPE :: tel. +31-70-3111010 http://www.pine.nl :: RIPE RegID nl.pine :: fax. +31-70-3111011 PGP key ID 92BB81D1 :: Dutch security news @ http://security.nl Today's excuse: the xy axis in the trackball is coordinated with the summer soltice From pgl at petrel.net Thu May 18 17:11:16 2000 From: pgl at petrel.net (Pascal Gloor) Date: Thu, 18 May 2000 17:11:16 +0200 Subject: RIPE Database crashes ... Message-ID: <22CF090CD80ED41189CE00E018C23A00E460@amonra.petrel.ch> Mark, sure, i will do it ! but RIPE need to give me access to mirror it :-) Pascal Gloor Network Manager _________________________ SPAN(tm) / Petrel Communications SA A Cable & Wireless company T?l : + 41 22 304 47 47 Fax : + 41 22 304 47 99 E-mail : pgl at cw.span.ch Web : http://www.span.ch > -----Original Message----- > From: Mark Lastdrager [SMTP:mark at pine.nl] > Sent: Thursday, May 18, 2000 5:04 PM > To: Pascal Gloor > Cc: lir-wg at ripe.net > Subject: Re: RIPE Database crashes ... > > At Thu, 18 May 2000, owner-lir-wg at ripe.net wrote: > > >What do you think about that ? > >Or what else solution do you have guys (LIR's and RIPE)... > > Well, maybe you can run a mirror database and give us the hostname? > > > Mark Lastdrager > Pine Internet > > -- > email: mark at lastdrager.nl :: ML1400-RIPE :: tel. +31-70-3111010 > http://www.pine.nl :: RIPE RegID nl.pine :: fax. +31-70-3111011 > PGP key ID 92BB81D1 :: Dutch security news @ http://security.nl > Today's excuse: the xy axis in the trackball is coordinated with the > summer soltice From andrei at ripe.net Thu May 18 18:22:39 2000 From: andrei at ripe.net (Andrei Robachevsky) Date: Thu, 18 May 2000 18:22:39 +0200 Subject: RIPE Database crashes ... References: <22CF090CD80ED41189CE00E018C23A00E45F@amonra.petrel.ch> Message-ID: <392418CE.76A2A2AC@ripe.net> Dear Pascal Gloor, Pascal Gloor wrote: > Dear LIR's, > > What's up with the database ? > Do you think it's normal that this machine crashes every 3-4 weeks ? > Does RIPE need some more guys to manage that ? > I think all LIR's whould be ok to pay RIPE a little bit more fee that RIPE > may have a VERY HIGH BIGGER MTBF (Mean Time Between Failure). > > What do you think about that ? > Or what else solution do you have guys (LIR's and RIPE)... The good news is that the Database is up and running again. The server crash did not affect queries, but delayed the updates. Nothing is lost and the updates are now being processed. To be honest, the last crash happened at the end of March, and that was a hardware problem. What happened this time was a break signal sent to DB server by the console server (which resulted in server reboot), the situation we have not expected, but sure it will not happen again. There is always a probability that something goes wrong, but we try to do our best to minimize the problems it causes. Regards, Andrei Robachevsky DB Group Manager RIPE NCC P.S. If you really plan to mirror the RIPE Database, you will need to sign an AUP. This is normal practice. Please write to to discuss this issue. > > > to be continued .... > > Regards, > > Pascal > PGL2-RIPE, AS6893, CH-PETREL > > Pascal Gloor > Network Manager > _________________________ > SPAN(tm) / Petrel Communications SA > A Cable & Wireless company > > T?l : + 41 22 304 47 47 > Fax : + 41 22 304 47 99 > E-mail : pgl at cw.span.ch > Web : http://www.span.ch From netmaster at space.net Fri May 19 09:08:53 2000 From: netmaster at space.net (Gert Doering, Netmaster) Date: Fri, 19 May 2000 09:08:53 +0200 Subject: RIPE Database crashes ... In-Reply-To: <392418CE.76A2A2AC@ripe.net>; from andrei@ripe.net on Thu, May 18, 2000 at 06:22:39PM +0200 References: <22CF090CD80ED41189CE00E018C23A00E45F@amonra.petrel.ch> <392418CE.76A2A2AC@ripe.net> Message-ID: <20000519090853.U2836@Space.Net> Hi, On Thu, May 18, 2000 at 06:22:39PM +0200, Andrei Robachevsky wrote: > There is always a probability that something goes wrong, but we try to do our > best to minimize the problems it causes. Servers just notice if you're out of country and find some way to fail behind your back... While everybody is complaining, I just want to note that the overall service of the RIPE DB is very good (the turnaround times for updates are slow during the day time, but knowing the history of the RIPE DB, it's amazing that it still works at all ;-) ). You are doing good work, RIPE-DBM people! greetings from Budapest, Gert Doering -- NetMaster -- SpaceNet GmbH Mail: netmaster at Space.Net Joseph-Dollinger-Bogen 14 Tel : +49-89-32356-0 80807 Muenchen Fax : +49-89-32356-299 From wh at bw.nextra.de Fri May 19 10:47:37 2000 From: wh at bw.nextra.de (Winfried Haug) Date: Fri, 19 May 2000 10:47:37 +0200 Subject: AW: RIPE Database crashes ... In-Reply-To: <22CF090CD80ED41189CE00E018C23A00E45F@amonra.petrel.ch> Message-ID: <000701bfc16e$e26e3560$fdd761c2@seicom.net> Hello, > -----Urspr|ngliche Nachricht----- > Von: owner-lir-wg at ripe.net [mailto:owner-lir-wg at ripe.net]Im Auftrag von > Pascal Gloor > Gesendet am: Donnerstag, 18. Mai 2000 16:51 > An: lir-wg at ripe.net > Betreff: RIPE Database crashes ... > > Dear LIR's, > > What's up with the database ? > Do you think it's normal that this machine crashes every 3-4 weeks ? > Does RIPE need some more guys to manage that ? > I think all LIR's whould be ok to pay RIPE a little bit more fee that RIPE > may have a VERY HIGH BIGGER MTBF (Mean Time Between Failure). > > What do you think about that ? > Or what else solution do you have guys (LIR's and RIPE)... > to be honest, RIPE is providing a really good job to the community. There are always reasons for faults, especially when human beings are involved :-) If something happens with the Database, there was no loss of data in the past years and if you submit a mail to RIPE it will be processed even if there is an outage. So you can trust on RIPE. If i compare the service of RIPE with other organisations we _have_ to work with, the service of RIPE NCC is nearly unbeatable. If you send an email to internic or denic you cant be sure that the mail will be processed and you must check every mail if it is processed or not and in case of faults nobody sends email like RIPE, that the service is interrupted and all mails will be queued. Even the desaster from this french-guy who deleted huge amounts of handles, took only one day for RIPE to present the tracking data to the community. So there are other organisations to blame before we start with RIPE - we think they are doing a really good job for all of us just my 0.02 -- http://www.nextra.de -- feel green-feel nextra -- Winfried.Haug at nextra.de - nextra Baden-Wuerttemberg | Winfried Haug Geschaeftsfuehrer Communication | Nadlerstrasse 21 D-70173 Stuttgart Service Provider GmbH | tel +49-711-6015-0 fax +49-711-6015-199 ---------------------------------------------------------------------------- -- From ripe-dbm at ripe.net Wed May 24 14:56:30 2000 From: ripe-dbm at ripe.net (RIPE Database Administration) Date: Wed, 24 May 2000 14:56:30 +0200 Subject: Delay in processing updates Message-ID: <200005241256.OAA26695@birch.ripe.net> Dear Colleagues, We have currenly a larger update queue than usual. We expect that updates in the queue will be processed in 4-5 hours. If you have not yet received an acknowledgement to your updates, there is no need to send them again. All updates received by auto-dbm at ripe.net have been queued. We are sorry about the inconvenience. If you have any questions, please contact . Regards, Engin Gunduz ____________________________ RIPE Database Administration. From stephenb at uk.uu.net Tue May 30 09:39:51 2000 From: stephenb at uk.uu.net (Stephen Burley) Date: Tue, 30 May 2000 07:39:51 +0000 Subject: Auditing tool Message-ID: <39337047.330F73D1@uk.uu.net> Hi I would just like to say the auditing tool is a great success and helps me fix problems before requesting more space thus saving lots of time and effort. I have one suggestion though, when the script shows the percentage used for our network it is still over 90% of the total used even with a /17 spare. This means we still would technically qualify for a new range of addresses ;). I was wondering could the script also show in real terms and not a percentage how much space is available in "/" notation. I know i am being a bit pedantic but its just so we do not hit the panic button when its not needed. Regards Stephen Burley UUNET EMEA Hostmaster From bochmann at freinet.de Wed May 31 16:54:23 2000 From: bochmann at freinet.de (Alexander Bochmann) Date: Wed, 31 May 2000 16:54:23 +0200 Subject: "ownership" on ancient network assignments? Message-ID: <20000531165423.E26359@freinet.de> Hi, we are currently noticing a dispute of two companies over a /16 address range assigned to a predecessor of both of them in 1991 (the original company was split some time ago). We're not a party in this quarrel, but I'm curious about the status of a network that was assigned such a (comparatively) long time ago. Can someone claim to be legally entitled to own this IP address range? What is likely to happen? A. Bochmann -- FreiNet Gesellschaft fuer Informationsdienste mbH Loerracher Strasse 5a D79115 Freiburg