[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [sbrim@cisco.com: Re: [RRG] Thoughts on the RRG/Routing Space Problem]

On 2007-12-04 06:09, Scott Brim wrote:

(another from wrong userid)

----- Forwarded message from Scott Brim <sbrim@cisco.com> -----

Date: Mon, 3 Dec 2007 06:41:49 -0800
From: Scott Brim <sbrim@cisco.com>
Subject: Re: [RRG] Thoughts on the RRG/Routing Space Problem
To: Brian E Carpenter <brian.e.carpenter@gmail.com>
Cc: Russ White <riw@cisco.com>, rrg@psg.com
X-Mailer: VM 8.0.5-504 under Emacs 22.1.1 (i386-apple-darwin8.10.1)

Excerpts from Brian E Carpenter at 15:03:37 +1300 on Mon  3 Dec 2007:

On 2007-12-02 02:24, Russ White wrote:
<snip>


2. An exponentially growing table of short prefixes, facilitated by the
huge IPv6 address space (as Tony has pointed out, routing /32's in IPv6
is the same thing as routing hosts in IPv4, so you've gained nothing in
table size if you go in that direction).

Er, yes, there's a reason that the IPv6 design assumes
provider-based aggregation of prefixes, and multiple prefixes if you
have multiple providers. The PI heresy is an import from IPv4
thinking. But if that heresy sweeps the world, we're going to need
LISP style mapping to exorcise it from the core. I think that's how
we got here.


So it's let's-have-fun-with-rhetoric day, eh?  We got here because of
the rate*state problems, of which PI allocations are a small part.
Once we got here, we discovered that we potentially had the freedom to
abandon the dogma that you are clinging to.  (At this point I believe
you're supposed to say "I find your lack of faith disturbing").


Oh, I do, certainly ;-)

However, my point was not intended as rhetoric. The reason we designed
IPv6 for multiple prefixes per site was precisely to avoid the problem
that IPv4 faced pre-CIDR. The emergence of PI allocations and BGP4-based
multihoming for a large number of IPv6 sites would recreate that problem.
It hasn't happened yet because we don't *have* a large number of IPv6
sites yet. But if we can't get people used to the idea of multiple
prefixes per site, we *will* have the problem, and I thought that
was the main reason we're here - to keep the number of prefixes that
the core has to route down to a manageable number, even if sites
stick to the old notion of one prefix per site.

   Brian

--
to unsubscribe send a message to rrg-request@psg.com with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg