Re: An idea: GxSE

["Jon (Taz) Mischo" <taz@tazlore.com>]

I understand your point, but with careful construction this shouldn't be
an issue.  Assuming you use a link state protocol for the interior portion
and you allow the border routers to notify the interior routers of the
egress point, you could move the rewrite closer to the edge.  In fact, if
you let the initial rewrite occur at the egress router, and then move the
rewrite backwards into the network, with the egress router indicating
validity of all recently used routes back into the IGP/other protocol for
this purpose, you would know where the packets are going.

Basically, if there is a topology change, you just need to make sure
your path is still valid.  Another behavior to control this may be to stop
rewriting at the edge if you detect a topology change that may affect your
internal routing, and let the egress router rewrite and re-alert you.

I had another point, but I forgot it...serves me right for stepping away
to deal with something else.

-Jon

On Tue, 26 Jun 2001, Ben Black wrote:

> On Tue, Jun 26, 2001 at 01:22:20PM -0500, Jon (Taz) Mischo wrote:
> > > Given the hop by hop nature of IP, I suspect it will be rather difficult
> > > to have the translations occur anywhere but at the border.
> > 
> > Not at all.  As I explained in a previous post, having the valid GR/SK
> > pairs announced in BGP or the like and propagating them through the
> > internal network via an IGP or a proptocols specifically designed for this
> > purpose would let you translate in the distribution layer or even access
> > layer (although in most cases the distribution layer routers would be the
> > ones beefy enough to handle it of the two).
> > 
> 
> This doesn't address the concern I noted above which is that the interior
> routers simply don't know the egress router, and hence the egress prefix
> (assuming you always map into space provided by a given upstream, which
> is how I would expect to avoid rapid table growth).  This information
> _could_ be communicated to the interior routers, but it would require them
> to carry full reachability information, which seems a rather onerous
> requirement.
> 
> 
> Ben
> 
> 

-- 
        "Be liberal in what you accept,
      and conservative in what you send."
--Jon Postel (1943-1998) RFC 1122, October 1989