On Fri, 15 Dec 2000, Geoff Huston wrote:
> Let me get this right..
Please do. :)
> Shane says "as for convergence times [...] I don't know" Does
> that need to be solved here?"
>
> well, you know, I thought the mission here was one of research into
> the chartacteristics of the space and thinking forward into potential
> approaches in a context of understanding what we need to know more
> about - i.e. research.
>
> I'd love to be so confident about convergence times and indicate that
> we know all there is to know about convergence in large dynamic and
> complex spaces. Even to know that some person on the planet has this
> knowledge would be comforting to some degree. I just don't see that we
> have this essential knowledge at present.
Am I the only one that got the _Everything I Needed to Know About
Convergence Times I Learned From Abha's Presentation_ book for Christmas
last year?
Seriously, why are you so hostile? I never said we know all there is to
know about convergence. In fact, all I said was: "I don't know." As in
"do NOT". As in NOT to know. As in, nearly the complete opposite of
knowing everything.
I've seen Abha's convergence presentation about 5 or 6 times now - and it
does get better as she gets more information. But for me personally, I
don't see how any system that requires as much human hand-holding as the
current Internet routing system (read: BGP) can continue to scale. It's
brittle. It's confusing. I often wonder how it can work at all (except
when my provider goes down, then I wonder more about the definition of
"work").
So, I guess I'll just patiently wait for the survey of current routing
research which is going to be dropped on the mailing list at some point.
That's a good thing - I certainly don't know who's doing what in routing,
or where, even if I had the time to follow up on it.
I am of course not saying that all experience gained is worthless, but I
think it's a shame to spend too much time examining the behavioral
characteristics of a system that should be replaced.
-- Shane "obviously not speaking for my employer here" Kerr
This archive was generated by hypermail 2b29 : Mon Aug 04 2003 - 04:10:04 EDT