Hello all, About the /32's; they are announced on purpose. The reason is that we have had problems with some ISP's BGP filters that are set to a maximum size of /32. As a result did not our IPv6 customers get connectivity via ISP's using the /32 size filters. Maybe this is a problem only we run in to being one of the first using a very large prefix. However, the BGP filters may be updated by now so we can stop announcing the /32's. Please note that this has been a temporary work around to maintain connectivity for our customers. We also have old prefixes that we are about to return to RIPE (2001:06C0::/35), but not yet had time to shout down (affects many customers and links). It is of-cause the policy of the TeliaSonera group to normally only announce one prefix. Any comments to this are most welcome. Best regards /// Mattias Lignell -------------------------------------------------------------- Mattias Lignell TeliaSonera Sweden, IP-Networks Phone: +46 8 713 3706 Mobile: +46 70 595 9068 Fax: +46 70 615 6245 Email: mattias.lignell@teliasonera.com Address: Vitsandsgatan 9, House-D, 123 86 Farsta, SWEDEN ICQ UIN: 18690075 --------------------------------------------------------------

------------------------------------------------------------- ----- Original Message ----- From: "Jeroen Massar" <jeroen@unfix.org> To: "Gert Doering" <gert@space.net>; "David Kessens" <david.kessens@nokia.com> Cc: "Amar Andersson" <amar@telia.net>; <ipv6-wg@ripe.net> Sent: Friday, September 30, 2005 3:06 PM Subject: Re: [ipv6-wg] Draft Agenda (v1) for ipv6 wg RIPE51

On Thu, 2005-09-29 at 20:46 -0700, David Kessens wrote: One discussion point you might want to raise: - what are people going to do with 6bone address space as per 6/6/6 ? - filter it? - keep it going? - warn people? In january I will send out a note to all the folks still announcing 6bone prefixes to remind them that they should be shutting it down per 6/6/6. There doesn't seem to be a large movement in doing so yet...

C. Discussion on: Global IPv6 routing table status (Gert Doering)

Maybe something related might be noted here, in the global routing tables I've seen for instance, just a 'random' targetted pick. http://www.sixxs.net/tools/grh/lg/?find=2001:2000::/20 [Notez bien: the question here is about what is going to become the common policy and what is going to happen to the ipv6 routing, it is not to 'attack' or critize an ISP on what they are doing] 2001:2000::/20 announced by AS1299 (TELIANET) 2001:2040::/32 AS3301 (TELIA-SWEDEN) 2001:2060::/32 AS1759 (SONERA-TRANSIT-AS) No route6 for these 3 prefixes and no inet6nums exist for the the two /32's. Which leads me to a couple of questions: - is it mandatory to register route6's and inet6num's or is it just 'being nice' - is the trend going to be that >/32's will be announced as separate /32's? This of course means that the ISP can be found under one block but still affects the number of routes in the routing table. One could filter the smaller ones if the parent route is present. The idea of minimizing routing table size (yes that is not an issue now, but behold the future) is totally gone when ISP's will do this. I do understand why: to make the branches independant and attract the traffic as local as possible. On the otherside there are also ISP's simply requesting multiple separate /32's from the various RIR's. eg UUNet: 2001:0600::/32 - Europe 2001:4440::/32 - Hongkong 2001:4441::/32 - Australia 2001:4442::/32 - Japan 2001:0588::/32 - South Africa (I am ignoring the 6bone block they still have, with 1 other already returned, as that one will disappear anyway) The question in case of this setup is, why didn't they get 1 block? The generic question: what is the correct plan for these setups?

D. Report(s) about *actual* v6 traffic volume as compared to v4? *what's real* out there, not what's on powerpoint? (input from the audience)

http://www.sixxs.net/misc/traffic/ :) Not much, but it is a bit and for the moment declining a bit weirdly enough. <SNIP>

For example, address fragmentation, a key problem in IPv4, degrades address lookup performance in routers. Thus, a well-designed address allocation policy needs to minimise fragmentation while using the address space efficiently.

See my related note above. Greets, Jeroen