Rob Blokzijl and David Williams, director of Computer and Networking at CERN welcomed the participants to the 10:th RIPE meeting.
The proposed agenda was approved. Logistics and the organisation for this meeting was decided.
Minutes from the 9:th RIPE meeting. CERN Networking Maps, July 1991. Report on East European networking, Aug 28 1991. Report from Joint Network Team (JNT), Sep 9, 1991. EARN NJE/IP links, Sep 1991. RIPE IP/IXI connections, Sep 20 1991. RIPE DNS Host Count, Sep 21 1991. RIPE European IP inventory Sep 23, 1991. RIPE Database report, Sep 24 1991. RIPE Routing WG report, Sep 26 1991.
2. Minutes and action list.
The minutes from the 9:th RIPE meeting was approved.
- Send out invitation to the four points on the envisaged RIPE backbone. CLOSED. (Due to the EBONE initiative).
- Distribute the RIPE NCC set up document to the RIPE mailing list for further comments. CLOSED.
- Approach international networking organisations in order to agree on a common position to share international lines leading to the central and east European countries and helping them improve their connectivity. CLOSED.
- Chase WG conveners to have the next session program ready one month before the next meeting. CLOSED.
- Contact CISCO to establish formal contacts. OPEN.
- Review documents to be installed at the nic.eu.net document archive. OPEN.
- Write a paper on policy based routing. CLOSED.
- Write and make easily available a cookbook about European network information services. OPEN.
Thomas Lenggenhager and Bernhard Stockman
- Write a document clarifying the meaning of "national IP infrastructure". CLOSED.
- Put together a list of locations where network maps can be retrieved. CLOSED.
- Set up for a routing group meeting. CLOSED. (Due to lack of time).
- Produce paper on the routing problem and how this could be solved. CLOSED. (Due to lack of time).
- Collect and document information on known routing restrictions. CLOSED. (Due to lack of time).
- Collect and document experiences in using the BGP routing protocol. CLOSED. (Due to lack of time).
- Progress the startup of a RIPE BGP pilot. CLOSED. (Due to lack of time).
Matti Rendahl. - Set up of European root name server within NORDUnet. CLOSED.
Milan Sterba - Write a summary of actual and planned international connectivity of Central and Eastern European countries and present it as a RIPE working document" CLOSED.
- Keep track on the development of ISDN connectivity in Europe. CLOSED.
- Set up, publish and maintain a list of persons responsible for international networking in the central and east European countries. CLOSED.
- Put information regarding national infrastructure in a presentational form. CLOSED.
- Coordinate the RIPE mapping efforts with corresponding IETF WGs. CLOSED.
- Send information on mapping topics to James Barr. CLOSED.
- Distribute the draft RIPE statistics document to the ripe-org mailing list for further comments. OPEN.
- Send a list of proposed but not installed documents to Rob Blokzijl. OPEN.
3. Short reports.
Olivier Martin reported on traffic statistics measured by CERN. (See separate paper). The report was followed by a discussion on redundant FTP traffic. This could be a subject for RIPE.
Phil Jones reported on the status of the JANET IP Service. (See separate paper).
Q: Who to contact in JANET with connectivity queries. A: Tony Bates or John Seymour, i.e. the contacts registered with RIPE for technical/operational queries.
Daniele Bovio reported from EARN. (See separate paper).
Ben Segal reported from the Trieste Networking School. A call for participation was announced and also the need for lecturers. An announcement will be distributed.
Antonio Blasco Bonito reported from GARR. An Italian registry is now installed. For more information regarding GARR use the email address mail-ita _at_ info.garr _dot_ it
4. RIPE Database Report (Daniel Karrenberg).
Daniel presented the latest database statistics (see separate paper). There has been a lot of work including correction of bad entries being sent for inclusion in the database. Daniel requested that entries should be checked with the corresponding scripts before being sent. Also these entries should only be sent to the ripe-dbm _at_ nic.eu _dot_ net mail address to be treated in due time.
5. EBONE-92 presentation. (Kees Neggers).
Kees Neggers presented the proposal for an interim 1992 European Backbone (EBONE-92) providing TCP/IP and ISO-IP services. The report has been distributed earlier to the ripe mailing list.
Frode Greisen gave a short report from the RARE Operational Unit Task Force. The task force gave recommendations for the Operational Unit:
- Shareholders from the participating networking organisations should be shareholders of the unit.
- It should be a not-for-profit entity. - Cross subsidies between services should be avoided.
- It should be customer driven.
- It should be funded by the shareholders.
- It will be governed by a board of directors representing the shareholders.
The Operational Unit will be discussed at the RARE CoA meeting in Rome, October 1991.
6. Report from the USA.
Elise Gerich, Merit. In the NSFnet T1 network SNMP and NNstat has been used for traffic measurements. Merit is now changing the way of gathering data as there are problems of uncounted packets when bursty traffic. In the T3 network there is no token ring where statistical programs could run in dedicated hosts on that ring. The new statistical programs will run directly in the routers. The presentation view-graphs is produced by a proprietary database tool (Style) and could not freely be distributed, but for those sites already having this tool Merits format descriptions could be distributed.
Steve Goldstein, NSF.
- The ICM-link to NORDUnet was recently upgraded to 256 Kbps.
- The ICM-link to INRIA (128 Kbps) is currently around 70% loaded.
- South African request for NSFnet connectivity is approved. South Africa are doing tremendous work to get neighbour countries connected.
- The CIX association have been announced. CIX-west is installed and CIX-east will soon be up and running. There has been a meeting to get a form for the CIX as we all shares a flat address space. The IETF Operations Area Requirements Directorate (ORAD) is a possible forum for these topics.
- SRI NIC has now been moved to GSI. NSF is planning a call for tender on behalf of the other agencies for a NIC. It could be the same contractor as for NSF, i.e. GSI.
- WAIS, Wide Area Information Service, is a tool public available from ftp-think.com. WAIS offers an advanced graphical interface similar to Archie.
- Export regulations for East Europe does not concern networks but information and equipment. Poland, Hungary and Czechoslovakia will have their request for connectivity to NSFnet approved. Further approval for e.g. Soviet Union will be investigated.
- Name space and address registration for CLNS is now being worked on in the US.
END of DAY 1 DAY 2 fully dedicated to WG meetings. During day 2 also the initial EBONE Consortium meeting was held at CERN.
7. Report from the EBONE Consortium meeting. (Kees Neggers)
The participants of the initial EBONE Consortium meeting were in general very positive to the proposal. Many representatives could not make commitments directly but would take this up in their organisation. A deadline for initial offers for contributions were set to October 30.
The EBONE Consortium of Contributing Organisations (ECCO) and the EBONE Action Team (EAT) were formed. The next ECCO meeting will be at October 30 and the EAT will have its first meeting directly after this.
8. The RIPE NCC progress report. (Rob Blokzijl)
No formal decision has yet been taken in RARE but there is good hope that this will happen at the RARE CoA meeting in Rome. An announcement for people will be made in the beginning of October. The RIPE NCC personnel will formally be employed by RARE until a dedicated legal entity is in place.
9. Reports from the RIPE WG sessions.
9.1 The Routing WG (Jean-Michel Jouanigot)
The distributed document was treated. The goal is to come to a recommendation for inter-regional policy based routing. One problem is that there currently does not exist any kind of published material regarding the routing policies among the RIPE connected networks. One possible implementation would be to enforce routing policies via routing centers. The routing policies have to be included in the RIPE database. Time schedule:
October, 7th : (Routing WG) Two new objects will be defined to be included in the RIPE database:
November, 1st: (All RIPE members) The routing policies of all the routing centers should be published in English and using the rules described in the routing document.
December, 1st: (Routing WG) New object/fields implemented in the RIPE database. Finalized definitions are checked and published by the working group.
December: (All RIPE members) Update the RIPE database information.
January, 1st: (All RIPE members) First use of the new objects/fields to implement routing policies.
A BGP pilot was proposed at the previous RIPE meeting but no progress so far. At the NORDUNET Engineering and Technical Forum (NETF) meeting next week, the possibility for such a pilot within NORDUnet will be investigated. Interested parties are invited to participate if this will happen.
BGP is really needed among the RIPE connected networks. The current CISCO implementation is believed not to be in accordance with the specification. (RFC 1163). CISCO should be contacted and informed about the requirements as needed within RIPE. A possible action could be to express these concerns in the Internet Monthly Report. Bernhard Stockman will take this as an action.
9.2 The Database WG session. (Daniel Karrenberg).
Due to lack of time there is still problems to update the database in a timely fashion. The possibility of remote updates was considered. Unauthorised access could however never be granted. A light-weight update procedure will be undertaken until the RIPE NCC is in place when more adequate procedure could be considered.
An possible improvement would be to decouple the updating of database information from the actual implementation of corresponding routing entries. NSFnet expressed their negative experience in having these procedures depending on each other.
Currently only the organisation running a network has the authority to register that network. Routing information should be sent together with the registration form for checking.
- A draft document will be produced on methods for incremental updates by Ruediger Volk.
- Another database tool distribution will be made available by Daniel Karrenberg.
- Contacts with other organisations will be made for coordination of databases (NIC coordination).
Final conclusion was that the RIPE NCC is very much needed to increase the responsiveness and accuracy of the database update procedures.
9.3 The East Europe Session (Milan Sterba)
An initial information session for new participants. A report have been distributed among the RIPE members but is already out of date. Milan will continue to update this document and make it available on nic.eu.net.
NSFnet connectivity issueds were treated.
A discussion on programs for network establishment took place. There exists currently plans for a Cseckoslovak backbone. There is a need for coordination among Poland, Hungary and Czechoslovakia.
9.4 DNS session (Francis Dupont)
There is a need for DNS statistics from the European root server at nic.nordu.net. Currently DNS query statistics is gathered but there could also be a need for measurements of the the DNS traffic on the US connections. Bernhard Stockman will inform Matti Rendahl, responsible for the root server.
A cookbook on DNS is badly needed. IETF User Services Area Council should be contacted. Francis and Milan promised to have a first version ready until the next RIPE meeting.
Report from the Yugoslavian network YUNAC DNS problems. (Marko Bonac). There has been problems with the Yugoslavian to level domain. Being member of the COSINE project a X.400 service was developed in YUNAC. IP services begun in small islands at the universities. Local departments tried to register as top level domains but YUNAC would rather see that corresponding regions and main sites were registered first and that the local departments were part of these domains. All parties have now agreed on a proposed solution. Marko is one of the responsibles for the Yugoslavian root server.
9.5 Statistics and Monitoring. (Bernhard Stockman).
The session started with a review of what kinds of tools actually being used within the RIPE community. Most commonly used tools as described were:
- CISCO accounting - PSI SNMP - DEC MCC - CISCO NetCentral - Cabeltron NMS - Home made SNMP tools. - Ping based logging and alerting.
A report from DFN on a performance test of IP/IXI with varying packet and window sizes was presented.
There is an need to integrate RIPE and EARN based statistics when EARN-II is implemented as this means that most of the international EARN traffic will run on top of IP.
There is a need for a small set of metrics and simple NMS tools to be used within RIPE.
Example of professional network reports could be seen from JvNCnet and Merit.
There was some concerns that statistical gathering should be undertaken locally but making the results public available. For this to work there is a need to identify formats and methods to be used. The IETF OPSTAT WG have done some work within this area which should be considered as input to the RIPE statistical efforts. Paul Bryant volunteered to write a proposal on what to collect and in which format to store.
Actions: - Distribution of the draft IETF OPSTAT documents. (Bernhard Stockman) - A document on raw data formats (James Barr) - A document on possible NNstat sites (Tony Bates) - A document on what to get and how to store. (Paul Bryant). - A NOCtools list for statistical tools
9.6 Mapping WG session (James Barr)
Mapping comes in some variations, handmade or automated. There is a need to identify which kind of maps are needed. James Barr proposed to dissolve the mapping WG and let it join with some other relevant group. There could also be a need for an explicit NIC/NOC-tools gathering WG which a mapping WG could be part of.
As there seems to be a need for a restructuring of the RIPE WG and Task Force structure (see also below) the RIPE chairman will contact some participants to further investigate the possibilities here.
9.7 Network Mangement. (Daniel Karrenberg)
The session started with a brainstorming around two topics:
- More information needed regarding network mangement tools. - Security management.
It was concluded that a RIPE Trouble Ticket system was not needed for the time being.
Security. RARE WG8 and the CERT system report. Should RIPE take active action, e.g. set up an own security emergency group. The RARE WG8 report will be carefully studied.
It was discovered that some RIPE connected sites does still not receive the CERT bulletins. Daniel Karrenberg will check.
Basic knowledge of security is sometimes missing, especially in newly connected networks. There is a need for recommendations in the security area. Erik-Jan Bos will produce a draft recommendations document.
NORDUnet is currently producing a checklist for NMS systems. The checklist will be discussed at the comming NETF meeting. Harri Salminen will input the list to RIPE.
Erik-Jan Bos promised to set up for demonstration of NMS systems at the next RIPE meeting. Anyone having something that could be demonstrated should contact Erik-Jan. As the RIPE schedule already is somewhat pressed this may create a need to extend the meeting to include such a demonstration.
9.8 Dial-up WG (Daniel Karrenberg)
A brainstorming session concentrating around two major subjects:
- Low cost serial lines.
- Dial up services.
Routing problems with dial up connections as the line is not up all the time. A recommendation would be needed here. A demonstration of the method and the equipment was suggested for the next RIPE meeting. Daniel will coordinate this. Interested should contact Daniel. Erik-Jan Bos will give a presentation of NetBlazers.
10. RIPE WG structure and RIPE NCC reporting. (Rob Blokzijl).
RIPE is growing and the NCC will be operational very soon. There is a need to reconsider the identity of RIPE. For example the WG areas needs to be redefined. There is a need for well defined areas with well defined chairs making things happen also between meetings.
The RIPE NCC will report to a group consisting of the RIPE chairman, the RIPE deputy chairman and the WG chairs. Requirements for participating in this group should be:
- Technical Specialist - Practical Experience - Based in Reality - Different backgrounds
Rob will write a proposal to be discussed on the RIPE mailing list.
A decision is foreseen at the next RIPE meeting. There is the need to find victims to progress work in the defined areas. There is also the need to have guarantees from the home organisation to set aside some time for the RIPE work.
The new RIPE IP inventory map has been distributed to this meeting. Connectivity is now expanding to new countries as have been shown in earlier reports at this meeting.
12. Report from other groups.
RARE. The decision for the RIPE NCC is foreseen at the comming RARE CoA meeting. There has been a proposal fro revaluating the RARE WG structure to get rid of non-relevant work. An investigation of the work currently being done within the RARE WGs showed that RIPE seems to be the only group reaching satisfactory results.
CCIRN Next meeting will be in Santa Fe. There are plans to invite Latin America representatives. APCCIRN/APEPG for Asia and Pacific is now in place. The EBONE proposal will be on the agenda.
IEPG/EEPG IEPG will have its next meeting in Santa Fe in conjunction with the CCIRN meeting in November 1991 just before the IETF meeting at the same location. IEPG is currently being chaired by Geoff Huston from AARnet in Australia. Co-chairs for North America is Elise Gerich, Merit, and for Europe Bernhard Stockman, NORDUnet.
The old EEPG group was dissolved after having fulfilled its task as given by RARE with the delivery of the EEPG report in May 1991. A new EEPG group will be created where the European IEPG delegation at the Paris IEPG meeting will form a nucleus. The objectives for the new EEPG will be in accordance with IEPG defined items of importance.
IETF There is an need for reporting from those participating at IETF meetings being funded by RIPE. No reports have been produced from the Atlanta IETF by Bernhard Stockman and Ruediger Volk.
Participants at the Santa Fe IETF applying for RIPE funding should send a letter to Rob Blokzijl describing their interests.
13. A.O.B No other business.
The meeting gave a grateful thanks to CERN representatives for a well and generously organised meeting. Rob Blokzijl thanked the participants.
- Produce a document on raw data formats for statistical data.
- Produce a document on possible NNstat sites among RIPE connected networks.
Tony Bates and Marten Terpstra
- Produce a document on experiences of running IP/IXI.
- Contact CISCO to establish formal contacts. - Review documents to be installed at the nic.eu.net document archive. - Investigate possible restructuring of the RIPE WGs and areas of interests. - Produce a document regarding the group to which the RIPE NCC should report back. The document should be distributed to the RIPE mailing list for discussions.
- Produce a recommendations document on security issues. - Coordinate preparations for a NMS demonstration at the next RIPE meeting - Prepare a NetBlazer demonstration at the next RIPE meeting.
- Produce a document on statistical metrics to gather and how to store this data.
Francis Dupont and Milan Sterba.
- Produce a cookbook on DNS. First version ready at the next RIPE meeting.
- Organize the policy based routing implementation for RIPE connected networks.
- Write and make easily available a cookbook about European network information services. - Check for RIPE connected areas not receiving CERT bulletins. - Make a new distribution of the database tools available. - Coordinate preparations for a Dial Up IP presentation at the next RIPE meeting.
- Distribute the NORDUnet NMS checklist to RIPE.
- Distribute the draft RIPE statistics document to the ripe-org mailing list for further comments.
- Send a list of proposed but not installed documents to Rob Blokzijl.
- Write a RIPE section to the Internet Monthly Report. - Inform Matti Rendahl on needs for DNS statistics.
- Distribute the IETF OPSTAT papers to the ripe stat mailing list. - Produce a report from the Atlanta IETF.
- Produce a document on methods for incremental updates of the RIPE database. - Produce a report from the Atlanta IETF.