Dear colleagues, please find below the minutes of our session during RIPE33 in Amsterdam. Apologies for getting this out so late - the delay is entirely up to myself. ---- Draft Minutes of Routing WG Session - RIPE 34, 22 September 1999 ---------------------------------------------------------------- Chair: Joachim Schmitz (JS395-RIPE) Scribe: Els Willems (EW21-RIPE), Ambrose Magee (AMRM1-RIPE) Attenders: 84 A. Preliminaries Joachim welcomed the participants, passed around the participants' list, and asked for a volunteer scribe. Els Willems volunteered to take the minutes and was supported by Ambrose Magee. The working group agreed upon the minutes of the previous meeting, and accepted the agenda circulated earlier on the mailing list. Review of previous actions: o 29.R1 G.Winters, J.Schmitz, RIPE NCC Definition of the IRR and an AUP -done- The definition and ideas on how to use the IRR are incorporated into the RIS and IRR reality checking projects o 31.R1 RIPE NCC, D.Kessens, J.Schmitz Basic design for an IPv6 IRR -postponed- Even though this topic is still considered important, there had been not enough input to continue o 32.R1 RIPE NCC, JLS.Damas Prepare draft document on RIPE-181 -> RPSL transition issues -in progress- The draft document is prepared, but not yet publicly available o 32.R2 J.Schmitz Write project definition for Routing Registry reality checking -done- B. Report from the RIPE NCC (JLS Damas) http://www.ripe.net/ripe/meetings/archive/ripe-34/presentations/routing-wg-r... l/index.html Three major topics were dealt with: * operational issues * software status * transition to RPSL Regarding operational information, the database still grows, and with an accelarated rate since beginning of the year. The most used objects by far are domain (52%) and person registrations (39%), while more important objects from the operational point of view are inetnum objects (8.3%) and route objects (0.7%). The biggest project currently running is the software reimplementation. Therefore, development of the old code has been frozen except for some bug fixes, performance tuning for some updates, and a newly introduced queue management software. Reimplementation has slowed down due to several incidents around the database. Still, the query part of the server is close to being finished. More details will be given in a later talk. The human interface of the database software now consists of three people: Marek Bukowy, Engin Gndnz, and Ambrose Magee. They are handling manually approximately a quarter of the roughly 1,600 messages per week. PGP deployment is making slow progress and has roughly doubled since last report. Regarding transition to RPSL, because it depends on the reimplementation there has been some delay. Thus RIPE is still at phase II. Some new commercial registries have started directly on RPSL using the perl version, which is also used by ARIN. RADB will migrate at the end of the year. At RIPE code development is supposed to be finished at the end of the year as well. This code will support core rps-auth functionality, and will work on RPSL only. RIPE-181 objects will be translated back from RPSL, and be kept for a limited time. This part of the transition will most likely begin early next year. An update will be given on the transition process at the Routing WG session during next RIPE meeting. C. Updates to RIPE-178 (C.Panigl) Christian received some comments on his paper on route flap damping (published as RIPE-178). The document was known by most of the people in the audience, and those who did not know it also did not operate backbones, meaning knowledge in the community about issues around route flap damping is sufficient. There has been a proposal to dispose off the "golden prefix" approach. Regardless of this discussion, the access lists described in RIPE-178 need to be updated. Comments and input on this matter should be sent to the mailing list until end of next week. Christian received the action to update document RIPE-178 accordingly. D. Multicast Routing The agenda had again the item "MBone Developments" on it, which was dropped because the interest on this topic was too low. However, in general Multicast Routing is still a ver important and interesting topic. Joachim asked the audience where multicast projects are running, and there were some answers, e.g. from Randy Bush (Verio), Marc Roger (Belnet), Wolfgang Tremmel (Xlink). However, they are not the only ones working on multicast. This topic is also new for public exchange points. Daniel Karrenberg pointed out that the AMS-IX will meet and discuss on Friday. Gerd Besch then described the multicast test bed project the LINX is planning. They will set up the multicast exchange on separate hardware first for testing. During the test period connections to are for free (approx until end of October). With corresponding experience multicast will become a regular offer by the LINX. Details are not yet settled. Unfortunately, the presentation original has not become available so far. During next RIPE meeting, multicast will be treated as a focal topic with several presentations. Joachim asked for contributions from the audience. E. Routing between Europe and Asia - An Issue? (the presentation will become available shortly on the RIPE NCC web server) Thinking about routing, we normally look at the topology of the network resulting from the peering/transit agreements. However, traffic of the Internet is running on physical links, which are affected by geography. Historically speaking, Internet and USA are synonyms, with the US being the birth place of the Internet, and considering size still by far forms its biggest part. Thus in its beginning the Internet was US centric with no direct connection between Europe and Asia who exchanged their traffic across the US. However, other regions are growing rapidly, and the former "satellites" Europe and Asia are developing their own structure. A closer look at some sources of data reveals that we are still US centric today. Examples are physical network connections (essentially submarine cables), host counts, and routing policies. These items were looked at in more detail. It is obvious that most capacity on submarine cables extends from or to the US. Also regarding hosts, most of them are located in the US. Routing policies registered in the IRR only contain very little direct connections between Europe and Asia. While we are still very much US centric we may think that routing between Europe and Asia is an issue. After all, going via the US means crossing 15 timezones compared to the "direct" connection of only 9 timezones (thinking of distance covered and number of systems in between). The number of routers and hops across the US is definitely bigger, and thus the question on general performance arises. Joachim then asked the audience about their impression and whether they consider this to be a problem. Looking at physical distance, the difference between both connections (Europe - US - Asia compared to Europe - Asia) becomes smaller because in the first case, cables are running on the Northern hemisphere, while for a direct connection Europe - Asia they extend around the equator, leading to roughly similar lengths. Christian Panigl and Randy Bush pointed out that the fact which connection is chosen very much depends on the economics. Prices are significantly lower in the US due to lots of competition. Moreover, there is no tier 1 telco outside the US, making the US again a center in networking. Another reason for the Internet begin US centric is that lingua franca of the Internet is English giving English countries a natural advantage, even though Randy Bush expects the number of hosts outside the US to exceed the US hosts in maybe 2 years. As a bottom line, the situation is not going to change, at least not soon. F. Routing Information Service RIS (A.Antony) http://www.ripe.net/ripencc/pub-services/np/Talks/9909_ripe34/ http://www.ripe.net/ripencc/pub-services/np/ris-index.html Antony Antony presented on the Routing Information Service (RIS) project which had been proposed in RIPE-197, and now been fully described in the new document RIPE-200. The project has its own homepage http://www.ripe.net/ris, and discussion is meant to be held on the mailing list of the Routing WG routing-wg@ripe.net. RIS collects default free BGP routing information from the worldwide routing table, and stores it over time. Thus, ISPs may use the data stored to analyze problems, or track usage of addresses and ASs. Randy Bush clarifies that this is not suitable for immediate debugging purposes, but for historical analysis which covers a completely different scope of problem analysis. An interactive interface will be provided to the community to query the resulting database, including flap statistics. Moreover, data collected will be the foundation for IRR reality checking (see following presentation). The RIPE NCC is performing this project because they are a neutral body, and the NCC already has practical experience from the test traffic project. The following part of the presentation discussed a prototype to gather additional experience, and a proposed final setup. They include distributed remote route collectors and a central data collector. The software will support the distributed approach, consisting of BGP listeners, data handling, and user interfaces. Current experience was outlined with expected data volume per peer, on raw data, and data to be stored in the database. It became obvious that this is an ambitious project. Jake Khuon pointed out that they are doing a similar thing (IPMA project). Thomas Schmidt asked which reports were to be expected, but Antony pointed out that they first need to gain some experience comprising input from the community as well. A first prototype for beta testing will most likely become available before RIPE35, and plans are to enter into regular production service in late 2000. Input from the community is requested. G. Reality Checking of the IRR (JLS Damas) http://www.ripe.net/ripe/meetings/archive/ripe-34/presentations/irr/index.ht... Looking at the Internet compared to its picture in the IRR, both seem to be disjoint. However, if real life data from the Internet is collected, it may be compared to the information stored in the database. This is essentially announced ASs, AS paths, and prefixes, a scope of data which is covered by the RIS. It is then interesting to find where there are deviations between the real Internet, and the IRR. This is most important to identify "hijacked" address space, or usage of IP address space or ASs which have not yet been allocated. The findings will be reported, and tools will be developed to help adapt data in the IRR to better depict real life Internet. But this is only one piece of the picture - at the same time, proper authentication methods for the IRR must be provided, so that only those parties may register data where they have authority. The project will start in January 2000. It is estimated that it will take 3 months time to develop reliable data acquisition and applications. The first report will most likely become available by RIPE36. Jake Khuon proposes to Joao to talk to Gerry Winters because much of groundwork was already done in the IPMA project (see their server at http://www.merit.edu/ipma). Y. General I/O with Other WGs There was no input from other working groups. The Routing WG will provide input to the IPv6 WG as RIS develops and is no longer restricted to IPv4. Z. AOB Since there were no additional comments or issues, Joachim closed the WG session. --- Agenda ------ R I P E 3 4 A M S T E R D A M Routing Working Group Session 22-September-1999 Draft Agenda A. Preliminaries (Joachim Schmitz) - introduction - participants' list - volunteering of scribe - agenda bashing - RIPE 33 minutes - actions from earlier meetings B. Report from the RIPE NCC (JLS Damas) C. Updates to RIPE-178 (C.Panigl) - input from the audience on route flap damping D. Multicast Routing - MBone Developments (N.N., to be confirmed) input from the audience - Multicast Routing Projects Introduction (J.Schmitz) LINX (G.Besch) Input from the audience E. Routing between Europe and Asia - An Issue? - A quick glance (J.Schmitz) - input from the audience F. Routing Information Service RIS (A.Antony) G. Reality Checking of the IRR (JLS Damas) Y. General I/O with Other WGs Z. AOB Summary of open Action Points ----------------------------- 31.R1 RIPE NCC, D.Kessens, J.Schmitz Basic design for an IPv6 IRR 32.R1 RIPE NCC, JLS.Damas Prepare draft document on RIPE-181 -> RPSL transition issues 34.R1 C.Panigl Provide updates to RIPE-178 Els Willems, Ambrose Magee, Joachim Schmitz, 15-Feb-2000 ----- Thanks Joachim Schmitz --- JS395-RIPE -- standard disclaimer ---