BCP: Slicing a /32 for an ISP

S.P.Zeidler spz at serpens.de
Mon Apr 14 11:33:27 CEST 2008


Thus wrote Marco d'Itri (md at Linux.IT):

> On Apr 14, Michael Adams <madams at netcologne.de> wrote:
> > I'm eager to hear more opinions about this. We are going to use
> > /64 for all kind of network interfaces including point-to-point
> > links and loopback interfaces.
> I recommend against this, because if you use longer prefixes then you
> can encode useful data and/or a hierarchy in the address.

... and likely break the KISS principle in the process.

You have about 65 thousand /64 per site, just how many point to point
links and LANs are you going to have? Is there something to be gained
in not using least surprise settings?

Also, if you do encode further information into your addresses, consider
ordinary network churn. Will you have to change all p2p addresses when
you add a new access router at a PoP, for example? Or change the existing
ones' model?

We are all well conditioned to preserving adresses and my brain goes funny
squeaky noises as well when I try not to do that, but with v6, it really
makes no sense to do so at the below /48 level.

spz at serpens.de (S.P.Zeidler)

More information about the ipv6-ops mailing list