Re: My planned work on networking stack

From: Gleb Smirnoff <glebius_at_cell.sick.ru>
Date: Tue, 2 Mar 2004 19:09:02 +0300
  Andre,

On Tue, Mar 02, 2004 at 03:34:54PM +0100, Andre Oppermann wrote:
A> > B> As to the second part of your mail: That sounds like a reasonable suggestion,
A> > B> I am sure Andre and others are paying attention to this and will take it on
A> > B> board when an implementation is nearer.
A> > 
A> > If this is OK from you, I start working on it (second variant using sockaddr_aspath).
A> > I'm willing to see this feature, and I have a good testing conditions for it.
A> 
A> Putting the AS path into the kernel routing table is certainly not the
A> right thing to do.  The kernel fib must be as small as possible.  All
A> information how a route made it there is pretty much irrelevant and
A> only the business of the routing protocol daemons.

  However, I can imagine at least 2 things, where we need AS pathes in kernel
(at least optionally).
  Nowadays, for continuing my work on bringing netflow implementation I need AS path
info in kernel. If we are planning to create routing solution based on FreeBSD,
we will need to support netflow. And only in-kernel flow catching can give us
high performance.
  In nearest future, you are going to implement policy routing. Are you going
to support constructions like: "match as-path XXX; set nexthop a.b.c.d"?

  I do not insist that AS pathes in kernel are good idea. If you show me an
other way to get AS information when constructing netflow exports in kernel,
I'd be thankful. I'd be also thankful if you describe how policy routing can be
implemented while no AS info in kernel.

  What do other FreeBSD networking withards think?

-- 
Totus tuus, Glebius.
GLEBIUS-RIPN GLEB-RIPE
Received on Tue Mar 02 2004 - 07:09:08 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:45 UTC