On Mon, 16 May 2011, Bjørn Mork wrote: > I don't think that will be much of a problem, given that you probably > have full control over the vserver network. Routing on longer prefixes > than /64 work just fine, and there are many advocating using /126 or > /127 for point to point links.
/126 is fine. /127 is not really a good idea beacause of DAD, and it is not like using /126 instead of /127 is going to waste too much valuable address space (unlike /31 versus /30 in IPv4). http://tools.ietf.org/html/rfc3627 However, don't go using non-globally-routeable IPv6 addresses on any interfaces where externally routeable traffic is going through. IPv6 really doesn't forgive ICMPv6 blackholes, and your router is likely to want to use the interface's IPv6 address to issue ICMPv6s. It is bad enough we have that kind of problem in IPv4. > I do find it extremely weird though, that they will provide you with 3 x > /24 IPv4 allocations but only a single /56 IPv6 allocation. Completely > insane. Not really. It is idiotic for sure, but not necessarily insane. Do note that RIR IPv6 address policies talk about issuing /56's to end-sites. But they never said anything about issuing just ONE /56 to an end-site. Still, it doesn't take too much effort to get that one wrong and confuse it with a recomendation of issuing a single /56 per 'end-site'. Add it to some 'cool' use of the IPv6 prefix to encode routing information, and you can easily get yourself tied to an extremely broken and stiff address plan that won't let you issue more than a /56 per datacenter rack. -- "One disk to rule them all, One disk to find them. One disk to bring them all and in the darkness grind them. In the Land of Redmond where the shadows lie." -- The Silicon Valley Tarot Henrique Holschuh -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected] Archive: http://lists.debian.org/[email protected]
