[85653] in North American Network Operators' Group
Re: shim6 (was Re: IPv6 news)
daemon@ATHENA.MIT.EDU (John Payne)
Sat Oct 15 00:06:36 2005
In-Reply-To: <Pine.LNX.4.62.0510141230310.26865@sokol.elan.net>
Cc: bmanning@vacation.karoshi.com, Joe Abley <jabley@isc.org>,
David Conrad <david.conrad@nominum.com>,
"Christopher L.Morrow" <christopher.morrow@mci.com>,
NANOG list <nanog@nanog.org>
From: John Payne <john@sackheads.org>
Date: Fri, 14 Oct 2005 23:50:26 -0400
To: william (at) elan.net <william@elan.net>
Errors-To: owner-nanog@merit.edu
On Oct 14, 2005, at 3:33 PM, william(at)elan.net wrote:
> No. The kicker is that the applications needs no such smarts and
> shim6 will take care of this for all applications on the system on
> the network level.
Not directly aimed at William: As others have said before (and I
finally listened) PLEASE take this discussion to the shim6 list.
However, I'm still at a loss as to how the shim on the end node is
going to know about the operational policy, cost implications and
site level load balancing requirements to do this _correctly_.
I'm also undecided about how I feel about the extra packets caused by
the (I forget the official term) discovery packets for shim6 for an
end site with say a hundred machines each with thousands of short
lived TCP sessions.
