On Jan 8, 2008, at 2:37 PM, Brian E Carpenter wrote:
I guess I believe that an unsustainable level of
BGP update traffic will be the first practical consequence to hit us.
The two router vendors that are vocally represented here don't seem
to support that view, however. How do you define "unsustainable"?
You can create non-convergence today in the lab by simply causing a
set of BGP speakers generate flap faster than the unit under
test. You can also create that in the field today by putting
woefully underpowered CPU (anyone still got a CSC-4 handy? ;-) in
the core. What we see today is that convergence is slowing, but
it's difficult to argue about when (or if) it will get to non-
convergence, given Moore's law and technology refresh cycles. This
is exactly why we try to focus the issue around clearly
quantifiable issues.
Tony