Thomas Lenggenhager welcomed the participants to ETH, Zuerich. The agenda was approved after some restructuring of the points. The time schedule for the meeting was settled.
2 Minutes from Last Meeting.
The minutes from the last RIPE meeting were partly complete, and handed out at the meeting. Some points needed additions or changes, which will be included in the complete minutes. Action Daniel Karrenberg.
3. Progress Reports from the Task Forces.
A new version (1.5) of the RIPE Task Force document is now available.
3.1 Task Force 1: Connectivity and Routing.
TF1-1. International IP Connectivity. A new version of the European IP connectivity map was handed out by Thomas Lenggenhager. The map is dated August 28, 1990.
A map showing the connectivity between Europe and the United States including also non IP connections made by Bernhard Stockman was handed out. This map is the first version and needs certainly to be improved. It was decided that changes and additions to the map should be communicated to Bernhard Stockman.
TF1-2. National not yet connected infrastructure. Information of most European national networks with both connected and not connected status is gathered. The material needs still to be put in a presentational form. Some countries like Belgium and Portugal are still missing.
Action Bernhard Stockman.
TF1-3. Plan for connectivity between RIPE and other continents.
TF1-4. Design of a routing scheme within RIPE. See agenda point 10 below.
TF1-5. Monitoring of RIPE routing. See agenda point 10 below.
3.2 Task Force 2: Network Management and Operations.
Within task force 2 subtasks 2 and 6 will have the highest priority.
TF2-1. Create and maintain a whois database. Version 2.0 of the RIPE whois database format is now available. The database now contains 1234 (!) entries with 599 persons and 635 networks of which 245 have RIPE connectivity, 170 have NSFnet connectivity and 189 have Internet Connected Status. NSFnet now has 1881 connected networks so RIPE is slightly less than 10% of that at present.The distribution of networks by country is:
The quality of the data has been significantly improved. The main source of inconsistencies now remains in the area of person/network relationships, like missing contact info or missing info about contact persons mentioned. It might be a good idea if the right person checked his/her entry in the database using the scripts contained in the database software package.
The database still contains no DNS info due to lack of software development time. This will be the next area to tackle.
Action Daniel Karrenberg.
Finally a remainder that updates should be sent to <ripe-dbm _at_ nic.eu _dot_ net> and not to Daniel Karrenberg personally to ensure timely handling.
TF2-2. Create infrastructure of operational contacts. Jean-Michel Jouanigot distributed a list of contact persons. This list will be improved.
Action Jean-Michel Jouanigot.
TF2-3. Procedure for notification when network unavailable. No formal progress but informal as most people know who to contact and how,hence there exists a network of informal contact ways.
Action Jean-Michel Jouanigot.
TF2-4. Procedures for common network operations. No progress.
Action Daniel Karrenberg.
TF2-6. Set up of an European NIC. See agenda point 7 below.
TF2-7. Maintenance and distribution of network management utilities. An improved version of the RIPE database programs and update tools is available by anonymous ftp from nic.eu.net in directory ripe, file ripe-db-sw.tar.Z. This contains scripts to check the consistency of (parts of) the database and to incorporate updates. It also includes a script to fetch all the preprocessed information that is contained in the ripe database to a local server.
Any database maintainer who wants to do distributed updates is welcome to propose a procedure. Once this is tried out it can become the recommended way for updates.
TF2-8. Create and maintain centres of expertise. No formal actions done yet. A lot of informal exchange is happening at RIPE meetings.
Waiting for someone to progress a more formal collection of information and distribution beyond the group that attends meetings.
3.3 Task Force 3: Domain Name System.
TF3-1. Creation of namespace database.
Heraklion domain, i.e. Greece is now here.
HEPnet physicists make efforts to get people in Soviet Union to communicate. We should stimulate these people to cooperate with each other. Surprisingly uptodate technology but missing infrastructure.
Soviet Union (.su) is supported by FUNET and CWI with secondary servers at normal places.
Soviet Union (.su) is being registered by the Unix User Group. Formal registration by the NIC is expected in the next few days. All interested parties within the SU may contact Dmitry Volodin. Dimitry's phone number is +7 095 231 2129. As soon as .SU is formally registered the mail address will change to dvv _at_ hq.demos _dot_ su.
Other domains now coming are Checkoslovakia (.cs), Poland (.pl) and Yugoslavia (.yu).
TF3-2. Coordination of backup name servers.
Yves Devillers is going to produce a paper on the domain data.
Action Yves Deviller.
Problems with subdomain nameservers that are not reachable. Subdomain nameservers should not be registered until they are surely working. At least one secondary subdomain nameserver should be globally available.
TF3-3. Design of a framework of DNS backup servers within RIPE.
The RFC's describing DNS are very general, no paper exists describing the European situation. TF3-3 will draft a document with a proposal for numbers and locations of European root name servers.
Action Francis Dupont.
There are problems with zonetransfers from INFN, due to software bugs in the Wollongong TCP/IP implementation. Problems with the .yu domain as their nameservers are used for dedicated X.400 conversions. DNS should be for everybody and not for a subgroup.
The root server within NORDUnet is not yet running due to technical reasons. We should really be concerned with the lack of root name servers within Europe. Matti Rehndal in NORDUnet is supposed to do some actions here.
Action Matti Rendahl.
TF3-4. Maintain list of DNS software and bugs.
A new version of BIND (4.8.3) is now available. Francis Dupont will start testing the release.
Action Francis Dupont.
TF3-5. Study mailrouting and MX records.
3.4 Task Force 4: Formal Coordination.
TF4-1. Template for IP cooperation agreements.
Action Mats Brunell.
TF4-2. Maintain contacts with CCIRN and FNC.
RIPE is invited to the EuroCCIRN meeting today (August 30, 1990).
See agenda point 4 below.
TF4-3. Start formal contacts with other organisations as necessary. Has been done as far as RIPE - RARE are concerned.
TF4-4. Propose legal status for RIPE.
As now being part of RARE, RIPE has legal status. This subtask is done and subtask 4-4 by this dropped.
3.5 Task Force 5: Statistics and Performance.
TF5-1 Development and evaluation of statistical tools.
The current statistics collected are in some points not relevant for the current network topology and the points where to collect the data should be reconsidered. For this reason a group should be formed to investigate points and periods for statistical data gathering.
Action Bernhard Stockman.
Tools for analysing longer periods of NNSTAT logs are now being developed and used within NORDUnet. If NNSTAT is to be used within RIPE these tools could as well be used there.
TF5-2 Presentation techniques for network performance statistics.
Ongoing development of the diagram tools. As soon as these changes are documented the programs will be made public available. TF5-3 Tools for benchmarking and performance evaluation.
3.6. Progress Reports on ACTIONs from the last RIPE Meeting.
1. Rob Blokzijl to request IP network number for IP connections over IXI.
Done! EUnet IP network number 192.16.192 will be used. Host numbers will be delivered by Rob Blokzijl or Daniel Karrenberg.
2. Eric Huizer to write a proposal on X.500 access to the RIPE Databases.
Eric Huizer or James Barr was not on this meeting. James Barr has written an interface from the whois database to X.500.
3. Daniel Karrenberg to progress task 2-4.
4. Jean-Michel Jouanigot to progress tasks 2-2 and 2-3.
Work on 2-2 has started and will continue.
5. Olivier Martin to make SAS programs for statistics reporting available.
6. Rob Blokzijl to contact Internet authorities about in-addr.arpa problem.
Has been solved due to RFC 1174 which sould be considered as an invitation for cooperation.
7. Rob Blokzijl follow formation of EEPG and liaise as necessary.
8. Rob Blokzijl to request RIPE travelling budget from RARE.
9. Daniel Karrenberg to prepare RIPE NIC requirements document together with Rob Blokzijl, Yves Devillers and Ruediger Volk to be ready at latest August 20.
Done! See agenda point 7 below.
10. Arnold Nipper to make input data for task 3-1 available to Daniel Karrenberg for inclusion in the RIPE Database.
11. Daniel Karrenberg to define RIPE DB format with long names, to revise output of whois server, to implement domain objects in the DB software and to add country designator to network objects.
4. RIPE and RARE, Status Reports.
Some money available from RARE for travelling. A budget shall be submitted to RARE concerning RIPE budget for the coming fiscal year (Jan 1 - Dec 31 1991). The proposed budget includes:
- Travels for one person to two CCIRN meetings per year.
- Travels for three persons to three IETF meetings per year.
- Travels to other important networking conferences.
- Secretarial support from RARE.
Report by Olivier Martin from EuroCCIRN meeting August 30, 1990. The meeting was mostly of informational nature. A lot of documents were handed out by Kees Neggers. A review was made of the action list. The representation in CCIRN was discussed. It was advised from CCIRN not to send more than 8 EuroCCIRN representatives. As there are fixed CCIRN repreresentation for the EuroCCIRN chairman, one CEC representative and the secretary general, it remains 5 people to be nominated. A list of 12-15 people were suggested of which 5 will be nominated by voting at the coming RARE CoA meeting. Because of the limitation to 8, it was thought essential by the RARE executive that the five to be nominated should not be appointed on a most important network basis but only on the basis of their personal expertise.
After this the report was discussed and the meeting decided to ask for a fixed representation of RIPE on CCIRN on the grounds that IP is the basis of a very significant part of intercontinental networking. The restrictions for those to be nominated mentioned above do not necessarily apply to RIPE as well, RIPE being not a network, but a body representing an important part of European networking.
Bernhard Stockman and Yves Devillers gave a short report from EEPG of which both now are members. The main topic for the EEPG is, in accordance with the tasks described by RARE CoA, to make a coarse analysis of the European networking situation to serve as input for technological and topological decisions concerning a pan-european backbone. Members in EEPG are nominated solely on the basis of their personal expertise and not on the basis of belonging to some networking organisation.
After this report it was decided that RIPE will help in giving input to the EEPG effort and also help in distributing information from the EEPG.
4.4. RIPE Deputy Chairman.
Daniel Karrenberg was unanimously elected RIPE deputy chairman.
5. Report from the FEPG and IETF meetings at Vancouver.
Rudiger Volk reported on these meetings.
FEPG met the day before the IETF meeting and European visitors were invited as observers. FIX points will be set up to interconnect federal networks. Currently two FIX points ready, FIX-east at the College Park NSS and FIX-west at the Palo Alto NSS. These FIX points will be the natural connection points for intercontinental links. However not all intercontinental links have to be connected to a FIX point. Eric Huizer gave a report fromthe EEPG which not could be considered as a counterpart to FEPG.
This meeting was the first IETF meeting outside US. IETF has now switched from 4 to 3 meetings per year. RFC 1174 was presented.Fergusson described the Canadian IP network which is hit by extremelyhigh PTT tariffs.
The PDN working group chaired by Rokitansky has been reformed with a new chairman to reflect all kinds of PDNs. SMDS will now be included in this working group.
Some confusion concerning IP network numbers, which are really connected and which are not but has Internet Connected Status (ICS). Around1700 networks are considered realistically connected while 3400 networks has ICS. There was also the problem of running out of IP address space.
Debate on IETF as a strictly technical body. Being part of standard development it should also consider itself as a standard making body.
Reports from some of the working groups.
Interconnectivity working group (IWG). Discussions on BGP version 2 which will not be included in CISCO 8.2. Also the BGP MIB was discussed.
Joint Management and Operation working group (JOMAN). One main topic was how to collect statistics. The possibility of having a small subset of the standard MIB defined to ease vendor implementations.
Topology and Engeneering working group (TEWG). Discussions on the supposed Europe - Japan link. Report from the Army Supercomputer Network. MILNET is a problem, thus there is the need of a more efficient network for military purposes. The T3 backbone will have separate and different NSSs. There is an interest in the european view of the transatlantic connectivity.
Next IETF meeting will be at NCAR in Boulder, Colorado, December 4-7, 1990.
After the reports from R"udiger Volkfollowed a discussion on how to organize RIPE representation at coming IETF meetings to cover relevant presentations and working group meetings.
Rob Blokzijl will expand the RIPE Task Force documents to be in conformance with corresponding IETF documents. This includes at least the name and email addresses of chairmans of the RIPE task forces. Those who wants to appear in this document should contact Rob Blokzijl. This document should after suggestion from Steve Goldstein be submitted as an RFC.
Action Rob Blokzijl.
Also the Ripe Terms Of Reference and the RIPE Requirements documents were suggested for submission. At the coming Interop90 a presentation of the European IP situation is scheduled. RIPE is questioned if anyone of us could do this presentation. Antonio Blasco Bonito and Juha Heinanen volonteerd and were accepted as RIPE representatives at Interop91. Material from earlier similar presentaions could be used if updated.
Action Antonio Blasco Bonito.
6. Report from NSF.
Steve Goldsteinnoted that he was very pleased to be welcomed as an obsever at the RIPE meeting, and that he was attending because of NSF's interest in supporting European plans for performing its own IP administration.
NSF considers the formation of a Network Coordination Center in Europe to be very important, not only with respect to globally distributed IP administration, but to offload some of the administrative burden of registering European IP networks from the NSFnet staff.
Concerning European traffic on the NSF backbone, NORDUnet accounts for more than 50(depending on the time of the sample). There are different organizations within US that administer the NORDUnet and INRIA lines which is a cause of problems. It would be more efficient if there was only one organisation managing the lines. A solicitation is for this reason send out. Latest date for incoming proposals is September 7. A NSF review panel will meet in Washington in September to evaluate the proposals. The hope is to have a signed agreement before the end of the year. Regarding NSF review panels in general, Steve invited people affiliated with RIPE to express interest in serving on NSF-Merit review panels dealing with relevant topics, in the future.
Three new NSSs are to be installed before the end of 1990 at Cambridge in Massachusettes, Atlanta in Georgia and at Argonne National Laboratory (ANL) near Chicago in Illinois. ANL will by this be directly connected to both NSFnet and ESnet.
A separate T3 nationwide network will commence with Phase 1, to be installed by the end of 1990. Phase 1 includes Cambridge, Ithaca,Pittsburgh, Ann Arbor, Champaign-Urbana, Argonne, San Diego and Palo Alto. Eventually, all of the NSFNET backbone nodes will be connected to the T3 nationwide network, which will then replace the T1 backbone. The new NSSs at Cambridge and ANL will intially be equipped as T3 NSSs. The current topological map of NSFnet as a series of NSSs connected by point to point links will no longer be adeqaute. With "core" NSS's located in the carrier's (MCI's) main junction centers, MCI could reroute traffic around a failing link as necessary. The adequate map will be NSSs connected to the MCI cloud with no explicit point to point links drawn.
The maybe first intercontinental CLNP connectivity test trial was set up between Ann Arbor and Bonn via Geneva. Steve Goldstein handed out the logs from an ISO traceroute and CLNP ping programs. GMD made available a topological map of the path of the CLNP ISO-grams. This trial supports the anticipated WG4 CLNS pilot under the leadership of Juha Heinanen. NSFNET plans to be a participant in that pilot.
As proposed in RFC 1174 connected status is decoupled from having an entry in the DNS database. It is necessary to fix the code in the DNS system and to enter several thousand formerly "un-connected" networks before this change will have become fully effective. So, in the meantime, we are still proceeding as in the past with using "connected status".
Internet Connected Status will be replaced by acceptable use statements and/or similar information to be included in new DNS data bases. One or more DNS data fields (replacing the field currently used for connected status) may in the future be used to store information regarding the tolerated transit traffic. This will enable the DNS to be used to support policy-based routing (for example, to insure that traffic does not transit networks which do not support transit traffic, or to exclude traffic from sources not authorized by individual networks).
In addition to RIPE and its efforts to implement a Networking Coordination Center in Europe, Canada's new national research network, CA*net has taken steps to perform its own IP administration. For Canada the University of Toronto will serve as the IP registry.
Earlier in July this year there was a meeting between NSF, DARPA end ESPRIT. Steve Goldstein had the impression that the DG XIII of the EC has not been kept informed of research and academic networking developments in Europe, especially as regards RIPE's activities and the advent of connectionless mode OSI services. Steve made it a point to brief senior DG XIII officials about such activities and U.S. - European cooperation toward an OSI "convergence" during private discussions.
A paper describing the convergence of United States and European networking is under preparation. This will be presented by Steve Goldstein and Christian Michau (CNRS/FR) at the IFIP WG 6.5 (MHS and upper level applications) meeting in Oerlikon, Switzerland. Two examples of US OSI activites. Larry Landweber, Universtiy of Wisconsin is starting X.400 tests within United States. He will also cooperate with RARE WG1. Alf Hansen (WG1, Norway) is on sabbatical and working with Larry to help implement X.400 service in the US Internet. PSI is doing an X.500 pilot. They are coordinating activities with UCL (Steve Kille) and others in Europe.
RIPE Documents produced and of general interest for the Internet network should be submitted as RFCs as the Internet according to Steve Goldstein is not just a North American affair but global. Documents like the RIPE Terms of Reference and the RIPE Requirements were suggested for submission.
7. Requirements for the Installation of a RIPE Network Coordination Centre (NCC).
A draft document was tabled desribing the requirements for a RIPE Network Coordination Centre that should be finalized in a form sendable to a funding body. The document describes all duties that has to be involved. This NCC shall not support end users but actual operation centres of the RIPE network. The main item is to collect and exchange information from local operation centres. Another important item is routing coordination. It must be very clear that this is NOT an operation centre but a coordination centre.
The document shall also refer to RFC 1174 as the Internet Connected Status now is dropped as an requirement in principle. This makes it necessary for a cooperation between a RIPE registry and other Internet Registries.
The document as proposed by Rob Blokzijl, Yves Devillers, Daniel Karrenberg and Rudiger Volk was handed out. This document is also available as ripe-19.txt on ftp.RIPE.net.
When finalized this document should be submitted as an RFC to give information to other networking organizations in similar matters.
No longer until the end of this year should the answer from any funding body be awaited. In this case the most natural body will be RARE. The RIPE NCC should be operational in January 1991.
The decision was :
- to revise the document according to the outcome of discussion within seven days.
- to have an email ballot with a period of seven days
- if the ballot concluded affirmative, then send the document to RARE with a request for funding.
NSF welcomes this RIPE activity and considers it of great importance to establish the NCC as quickly as possible.
8. Report from RARE WG4: The ISO IP Pilot Project.
Juha Heinanen will send in a proposal to RARE for the funding of the ISO IP pilot project. The participants will be CERN, NSFnet, NORDUNET, SWITCH and SURFnet.
A paper describing the review of ISO IP will be included together with a paper from Piet Bovenga, chair of WG4. See Apendix B and C.
9. Cooperation Between Europe and United States.
There should be better technical relations between RIPE Task Forces and IETF. One action is the above mentioned restructuring of the RIPE Task Force document to make it easier for IETF people to find contact points.
RIPE Documents that could be considered having common interest for the Internet communtity should be submitted as RFCs. So far is the following documents suggested for submission.
- RIPE Task Force document (restructured).
- RIPE Terms of Reference.
- RIPE Requirements document.
- RIPE NCC Requirements document. (Leaving out funding parts).
- RIPE Root Nameserver document. (To be produced).
Popular archives should be redistributed to US, Pacific and Europe. Some archives are already distributed from US to Europe like the NRI internetdrafs and IETF archives. For other archives more work is necessary.
At an informal meeting during the Florida IETF meeting some actions was mentioned. Bernhard Stockman will send out this list to relevant people in RIPE. ACTION Bernhard Stockman.
RIPE could be represented when relevant in NSF review panels.
10. RIPE Routing Matters, Including IXI and EASI.
Report from the routing group. IGRP could be used as a routing protocol for the RIPE backbone. Any CISCO could be part of this backbone. Non RIPE connected networks should be filterd out at gateways connected to the RIPE backbone. As this will not guarantee for protection against errounous announced networks, also some kind of consistency check must be performed. For this reason there must be a database and software tools to make this consistency check easy.
A prototype of the software tools is available as part of the RIPE Database software package. The database does not contain all necessary information yet. Daniel Karrenberg will coordinate writing a document about the filtering and the consistency check as part of task 2-4. SNMP access, at least ReadOnly to every gateway connected to the RIPE backbone must be granted. Some community string could if necessary be defined.
The backbone shall be one Autonomous System. Other Autonomous Systems will enter the backbone using EGP.
Third party traffic could be allowed.
IXI should be viewed as a point to point connection. I.e. IXI can not be part of the RIPE backbone. IXI should when used only be entered once. The actual routing using IXI is to be decided on within Task Force 1. Any host connecting to RIPE via IXI shall use the dedicated IXI IP network number. Host numbers will be given by Rob Blokzijl or Daniel Karrenberg as mentioned above.
EASInet is viewed as an integral part of RIPE. A draft proposal consisting of two documents will be circulated for comments.
Action Anders Hillbo.
11. New and Upgraded IP Links.
CWI - Falls Church. The link will be upgraded to 128 Kbps or more before the end of the year. Money is already allocated for this purpose.The INRIA and NORDUnet backup agreemnets remains as before.
CWI - UKC. Upgrade to 64 Kbps is ordered.
CWI - INRIA. Upgrade to 64 Kbps is ordered.
CWI - UNIDO. Upgrade to 64 Kbps is ordered.
CWI - NORDUnet. Due to problems with CISCO agent in Sweden the upgrade of the NORDUnet - CWI link is delayed as is the switch to IP on this link.
11.2. NIKHEF - CERN link.
No progress but converging discussions. At the moment it will probably be EASInet, EUNET, HEP and SURFnet that share the lines. The bandwidth will be at least 256 Kbps. There will be a meeting September 29 with the aimed outcome of a agreement between the parties.
11.3. Europe - Pacific connection.
Since Mats Brunell sent out his proposal for cost sharing nothing more seems to have happend.
11.4. UK - NSFnet fat pipe.
The link is now operational.
11.5. INRIA/Rouquencourt - INRIA/Sophia.
Upgrade to 128 Kbps ordered.
11.6. CERN - Vienna.
The link is now operational at 64 Kbps.
11.7. CERN - DESY
Ordered link with 768 Kbps.
11.8. Germany - NSFnet fat pipe.
No link ordered yet.
11.9. INRIA - SWITCH.
Backup agreement for CERN between INRIA and SWITCH has been discussed. SWITCH now accepts third party traffic.
11.10. NORDUnet - US.
Link is now changed to 64 Kbps terrestrial.
12. Closing of the 6th RIPE Meeting.
The participants thanked Thomas Lenggenhager and SWITCH for a well arranged meeting.
13. Action List
To complete the minutes from the 5th RIPE Meeting.
Inclusion of DNS data in the whois database.
Presentation of current national connectivity.
Startup of group for evaluation of statistics.
Circulate old material concerning cooperation Europe - US.
Produce a paper concerning DNS data.
Produce a proposal for European backup root name servers.
Testing of the BIND version 4.8.3 release
Set up of the root name server within NORDUnet.
Finalizing of IP cooperations agreement document.
Write a proposal for X.500 access to the RIPE whois database.
Progress TF2-2 and TF2-3
Make SAS program available for statistical reports.
Expand the current RIPE Task Force document.
Antonio Blasco Bonito and Juha Heinanen.
Presentation of European IP situation at Interop90.
Producing proposal documents concerning RIPE routing strategy.