[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
stla registry db issue
On Thu, 23 Dec 1999, Bill Manning wrote:
> % > so what are we supposed to do with the /35 sTLA allocations that
> % > the RIRs dish out for permanent (as opposed to 6bone) addresses?
> % The RIR policy is to reserve the whole /29 for expansion of the /35.
> % They promised never to allocate parts of the same /29 to more than
> % one user.
> % Therefore, the /35 can be announced simply by announcing the whole /29
> % that contains it. There will never be anyone else in that /29, so there
> % is no need to announce the longer prefix.
> % We must establish this principle solidly now, so that we **never** see
> % holes punched in /29s.
> % Brian
> This has its own set of problems. See the current discussion
> on micro-allocations inside ARIN. In IPv4 parlence, it seems
> a tremendous waste of space to delegate a /19 for a site that
> will never have more than a few hosts yet will be multiply
> homed. What I see here is the same argument, a /29 for a small
> set of nodes that will -never- meet the growth prospects for
> numbers of end-nodes. Historically this was also the case for
> IPv4 and its design, pre-subnetting. It became clear that
> there would not be a small number of networks with millions of
> nodes. Hence the initial subnettting model (class A/B/C) and
> the CIDR refinements.
> Is it just me or are we failing to learn from history here?
> This looks like a small number of networks with order(n) number
> of end-systems... all over again.
A micro allocation in V6 is quite a bit different to V4 in that the basic /64
unit should cater for virtually unlimited numbers of hosts. Allowing a small
number of bits for subnetting & routing would be all that's needed, so I would
suggest anything between /48 & /64 would be more than adequate.
Peter R. Tattam [email protected]
Managing Director, Trumpet Software International Pty Ltd
Hobart, Australia, Ph. +61-3-6245-0220, Fax +61-3-62450210