RIPE 50

• content to the Chair of the working group.
• format to webmaster@ripe.net.

RIPE NCC Services Working Group Minutes

Date: Thursday, 5 May 2005
Time: 16:00 - 18:30
Room: Plenary Room

A. Administrative Matters

* Welcome
* Select a Scribe
* Distribute Participants List
* Finalise Agenda
* Approve Minutes:
http://www.ripe.net/ripe/wg/ncc-services/r49-minutes.html

B. Report from the RIPE NCC - Axel Pawlik
http://www.ripe.net/ripe/meetings/ripe-50/presentations/ripe50-serv-ncc-new.pdf

Axel noted that the reporting on the RIPE NCC would not be repeated in the General Meeting scheduled to take place at 17:00.

• Registration Services
• Training
• Engineering
• New Projects
• Coordination
• Communications
• Membership Relations
• External Relations

There were no questions.

C. New LIR Survey - Results and Feedback - Paul Rendek (RIPE NCC)
http://www.ripe.net/ripe/meetings/ripe-50/presentations/ripe50-services-new-lir.pdf

There were no questions.

D. A New Version of the Hostcount - RIPE NCC

Timur requests feedback on the value of this service.

Questions:
Kurtis asks how many are using this Hostcount data? A few people raise their hand.
Does anyone have any comments about the proposed changes?

Lars-Johan Liman - Netnod:
Are you going to use this host query all over?

Timur: Yes.

Lars-Johan Liman - Netnod: Because I noticed some problems with some transfers. So it should be configured to provide one answer instead of many answers.... so this is about the zone transfer format.

Timur: Thank you for this. We are going to review all the projects and will maybe modify the Hostcount programme itself and maybe we should do a DNS as a backend, which may be better.

Peter Koch: I want to respond on the remark made by the previous gentleman. The host programme is showing its age. The sad fact is that the author passed away a couple of years ago and the program itself has been orphaned. There are some problems in dealing with the new style referrals. It would be useful to have that located at a well-known place, one of which could be the RIPE NCC. And I would be willing to contribute. As before I didn't do the coding, but it would be useful to develop that not only for the host count, but for DNS debugging as well.

Timur: Well, we promised to provide a DIY kit. We will come with a replacement on modified version of software. And your input would be much appreciated.

Souissi Mohsen - AFNIC: I browsed through the site and randomly checked some ccTLD's to find out about things. I was pleasantly surprised that there is detail on every ccTLD. But I am quite worried about the relevance of these figures. For example, the zone count.

If these figures are to be used somewhere, maybe there is a column missing - the estimate of error. Is there mathematical way to estimate that? And write a disclaimer for people using these figures?

Timur: Well, there is a disclaimer there,

Souissi Mohsen - AFNIC: Because the figures are, in some parts, very far from the reality

Timur: We understand that, so that's why we are not claiming that it's something really precise.

Souissi Mohsen - AFNIC: Thank you anyway.

Rob Blokzijl: By the way, the first Hostcount counted 7200 hosts. But that was never published.

Comment: I was one of the few who raised their hand when you asked whether people were using it or whether it is useful. I think it's extremely useful because it is one of the few long-term efforts to measure something of the Internet in a consistent way. It is more of a trend measurement than absolutely numbers. I would like to see some sort of continuation of this and I think its important that potential users understand what it is. What it is that you measure, what the conditions are and yes, its difficult to give error estimates, but people should use it as one of the few reliable trend figures. You can see trends in many publications, but most are based on nothing and if you take any given moment in time, you have a rich choice of various trends, like the internet is not growing anymore or the internet is exploding but at least this measures something that is defined and is not a Hostcount as the name suggests, but you are measuring something in a consistent way for many years, so you can draw some conclusions and I think that is the value.

Person from Vienna: I wholly support this project and I am very pleased to see that there is further work being done on this. I would just like to ask if, for the reverse DNS counts may LIR's volunteer to reform the divulged counts in the same way that some of us volunteered to perform forward counts for the country codes?

Timur: That's perfect. We would like to see participation. Thanks a lot.

Bruce Campbell: with what you said about the accuracy - if you see the graph on the screen there is a section that says zones blocked etc. That is the closest that you will get to seeing how inaccurate the figures are, because for every zone where we got a block, we have a completely unknown number of hosts that we couldn't count, which is why we started collecting the reverse count because its much easier to enumerate through the reverse zones and not necessarily forward. The downside is that people generally don't populate the reverse zones. So if you have reverse zones, it's putting them in DNS is good. It's yet another trend. It's purely a trend thing with no relation to reality.

Timur: Thank you

Peter Koch - DENIC (one of the starting points into host zone and long time support and I appreciate the effort and would like to see it continue). I appreciate your new considerations of acceptable use and really, there is no acceptable use. The raw data is no longer shipped, which is probably one of the pre-conditions for the ccTLD registries to continue support for this project. This is really important and we should talk about that.

Timur: Thank you.

Out of time, so no open microphone.

Y. Open Microphone

Z. AOB