"Jason Schiller ([EMAIL PROTECTED])" <[EMAIL PROTECTED]> wrote:
|On Fri, 14 Oct 2005, Dan Lanciani wrote:
|
|[snip]
|>
|> IPv6 itself does nothing good for (3). PI allocations may well be available
|> to some set of entities for a while to ease the transition, but that just
|> brings us back to routing table concerns. There is no general PI solution
|> on the horizon, and shim6 may make such a solution less likely to appear.
|>
|> Dan Lanciani
|> [EMAIL PROTECTED]
|>
|Dan,
|You clearly state that PI allocations do not help the concern that
|allowing de-aggregates in the IPv6 routing table could easily scale beyond
|current hardware limitations.
I think PI allocations *create* the scaling concern.
|I think this is what you are referring
|to when you say "There is no general PI solution on the horizon..."
Certainly I'd settle for a PI solution that does not involve putting
the PIs in a central routing table. If (as seems to be the case) we
can't even discuss distributed routing tables that appears to require
identifier/locator separation.
|But I want to point out that ARIN does have an IPv6 PI policy proposal on
|the table for the end of October.
Yes, that's what I meant about PI being available to some entities to ease
the transition and preserve the status quo. For a "general solution" I'd
expect availability to anyone who wants PI, and at trivial cost. (Cost
includes cost of getting the prefix advertised.)
|This policy does not provide a solution
|to routing table growth (possibly people believe everyone's hardware will
|be capable when the time comes that we need it).
No, people believe that PI space will be confined to large corporations who
can afford the cost. Remember, we've been through this before with IPv4.
Provider-allocated aggregated addressing was supposed to be a temporary hack
while the hardware caught up with routing table size. The hardware caught
up a long time ago but we never went back to generally-available PI addresses.
Instead the nature of the problem somehow changed from storage space to
computation or bandwidth or something else too nebulous to ever be solved by
hardware.
|If the ARIN IPv6 PI policy gets accepted and widely implemented that PI
|addresses are more likely to be leveraged by large commercial networks who
|are currently doing traditional IPv4 multi-homing and traffic engineering
|instead of shim6. This may greatly decrease the support for and use of a
|shim6 solution.
Yes, though it all depends on just how available those PI allocations are.
|It seems to me that if PI has wide spread adoption then everyone will have
|to commit to solving the large routing table problem in hardware. If we
|have already decided to solve the large routing table size in hardware
|then why do we need shim6?
I'm not at all convinced that we need shim6 (or even that it is desirable),
but as far as I know nobody has committed to solving the large routing table
problem in hardware or otherwise.
Dan Lanciani
[EMAIL PROTECTED]
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
