Hey, you nailed me on the head!<div><br></div><div>I have a /48 from ARIN and tried to advertise a /51 in datacenter 1 and a /51 in datacenter 2. Upstream, in massive quantity, rejected all advertisements < /48.</div><div>
<br></div><div>Which, unfortunately, is a problem. That means if I want to use my own address space in two datacenters, I have to go get another /48 for the other datacenter. This will indeed grow the routing table, when I have no need for so much address space.</div>
<div><br></div><div>Scott<br><br><div class="gmail_quote">On Wed, May 12, 2010 at 4:31 AM, George B. <span dir="ltr"><<a href="mailto:georgeb@gmail.com">georgeb@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div class="im">On Wed, May 12, 2010 at 12:50 AM, Rolf Hanßen <<a href="mailto:nsp@rhanssen.de">nsp@rhanssen.de</a>> wrote:<br>
> Hi,<br>
><br>
> it looks to me you calculate with an amount of v6 routes similar to the<br>
> number of v4 routes today and I thinks that this won't happen.<br>
> For example: RIPE has assigned 5x IPv4 PA networks to us during the last<br>
> 10 years but we only use one /32 v6 network. We furthermore had to splitt<br>
> up those v4 assignments because we have a split network with several<br>
> independent locations not connected together. This increased the number of<br>
> prefixes we announce very fast.<br>
> So the amount of v6 prefixes created from us will be much lower and I<br>
> think other providers may have similar spaces they use.<br>
> Even if you enable v6 for all networks currently using v4 I think total<br>
> amount will keep below 100k routes.<br>
><br>
> kind regards<br>
> Rolf<br>
<br>
</div>For people who can qualify for a /32, particularly those who grew late<br>
in the game, yes, there will be considerable consolidation. What I<br>
was more concerned about was smaller end user networks who currently<br>
have a multi-homed /20 or smaller allocation that will be getting a<br>
/44 or smaller v6 allocation and will keep both the v4 and v6<br>
addresses.<br>
<br>
There are a lot of multi-homed /24 nets out there that will become a<br>
/24 v4 and a /48 v6 announcement. I suppose the vendors could help<br>
things somewhat by not putting the entire 128 bits into hardware.<br>
There is really no reason to have more than 64 bits for routing as the<br>
last 64 bits are "supposed" to be host IPs ... but how many out there<br>
are using /127 for point-to-points between routers? You aren't<br>
"supposed" to subnet anything smaller than a /64 but many do.<br>
<br>
One thing I have noticed is that some networks are apparently<br>
filtering anything smaller than a /32 from PA space but allowing<br>
smaller nets from PI space designed for that purpose (down to a /48, I<br>
think). Anyone trying to multi-home a /64 is going to have a hard go<br>
of it, I think, but someone is bound to try!<br>
<div><div></div><div class="h5"><br>
George<br>
<br>
_______________________________________________<br>
foundry-nsp mailing list<br>
<a href="mailto:foundry-nsp@puck.nether.net">foundry-nsp@puck.nether.net</a><br>
<a href="http://puck.nether.net/mailman/listinfo/foundry-nsp" target="_blank">http://puck.nether.net/mailman/listinfo/foundry-nsp</a><br>
</div></div></blockquote></div><br></div>