Global Routing Operations (grow) Working Group MONDAY, March 20, 2006 1300-1500 (Afternoon Session I) ====================================================== CHAIR: Geoff Huston AGENDA o Administrivia 5 minutes - Mailing list: majordomo@lists.uoregon.edu subscribe grow - Scribe (text and jabber)? - Blue Sheets o Agenda Bashing 5 minutes Huston o Review and status of work items Active Drafts ------------- draft-ietf-grow-rfc1519bis-04.txt 5 minutes Fuller/all draft-ietf-grow-anycast-03.txt 5 minutes Abley/Lindqvist draft-ietf-grow-mrt-01.txt 5 minutes Blunk o WG Adoption of draft-scudder-bmp-00.txt 10 minutes Dave Ward (for John Scudder) o Routing Scaling Scenarios - some open questions Vince Fuller 20 minutes o IPv6 Potential Routing Table Size Jason Schiller 20 minutes o A look at BGP across 2005 Geoff Huston 20 minutes ===================== NOTES o WG Update Dave Meyer has resigned as GROW WG co-chair. The WG expressed its appreciation to Dave for his support of GROW since its chartering. o Review and status of work items draft-ietf-grow-rfc1519bis-02.txt BCP. In IESG evaluation, and outstanding DISCUSS tokens being considered. draft-ietf-grow-anycast-03.txt WGLC completed 22/11 Further revision made as a followup to AD comments. The WG was asked whether there was interest in performing a second WG Last Call for the document. No interest was expressed in performing a second WG review and the chair will pass the revised document to the AD with a publication request. draft-ietf-grow-mrt-01.txt Ready for a WG Last Call? The WG was ready for this document to be WG last called. The document editor will be consulted on this. o WG Adoption of draft-scudder-bmp-00.txt Following from consideration of this document at IETF 65, the WG was asked whether they wished to adopt this draft as a WG draft. A question was asked relating to the use of MRT in BMP, and it was accepted as a reasonable revision to the draft to accept MRT, as the differences are small enough to use MRT. The draft will be revised to include MRT and resubmitted as a WG draft. o Routing Scaling Scenarios - some open questions o IPv6 Potential Routing Table Size o A look at BGP across 2005 Dave Meyer reported to the WG that the IAB is to form a design team to take a look at routing and address scaling. The WG was asked to email iab@ietf.org with ideas. Vince Fuller: presentation on open issues with ipv6 routing/multi-homing. Questions and Comments: o Does address and topology have to be isomorphic? Quoting Yakov Rekhter, "Addressing can follow topology or topology can follow addressing. Choose one." o When you aggregate you throw away information. When you aggregate you get benefits. When you deaggregate you expose the dynamic nature of the system. o Moore's Law is an observation. Not a rule. o 8+8 and GSE were apparently not further progressed for political reason.. can anyone elaborate? o There were perceived security issues, surrounding the perception that ignoring the upper 8 bytes of address opens up some possible attack vectors. o Read the ESD analysis - recommended reading. o Comment that 8+8 was an innovative approach, but there evidently are problems with 8+8 that could not be solved. The result was from direct input. o Comment that 8+8 was the sole topic for ipng meeting, number of open issues. ESD was technical analysis, share common concern to routing stability. It changes dependencies. ie DNS circular security is an major issue. o IPSEC installed on all hosts would mitigate security issues. Jason Schiller: presentation on potential routing table size Questions and Comments: o If you assume widespread adoption, do you fix the igp before the egp? o No, this presentation is not looking at the igp, the igp issue could be an order of magnitude larger. o Why didn't you bring in vpn routes? o This exercise was just looking at internet routes. No clear good data to investigate, again issue could be much larger with the vpn routes included but data is variable and difficult to project. o What about the tax on control plane? and number of paths to investigate? o Can consider that as well as spf and fib/rib relationship. Not always directed at the amount of ram but the cpu has impact. Geoff Huston: presentation on potential routing table size o it was noted that this was an eBGP view o There was discussion of whether the MTU was relevant in prefixes per update - some further investigation of this could be helpful o The question was asked about implicit withdraws - and it was noted that the load of implicit withdrawals was not included in this study. o A question was asked about flap damping and it was noted that it appears that many transit ISPS were not doing it, given the obvious flap levels of certain prefixes. It was reported that less that half from nanog survey said they had damping on.