Mailing List Archive

On Wed, Mar 09, 2005 at 10:02:38AM -1000, Randy Bush wrote:
[snip]
> *extremely* initial results, which have now known but minor
> errors, were presented at apnic. see
>
> <http://rip.psg.com/~randy/050223.anycast-apnic.pdf>
>
> better analysis will be available in a week.

Never one to grump at free & meaningful info, I didn't beat
the drum on the 'week' marker. Just wondering if there is a
horizon to the more detailed/better analysis.

> imiho, the bottom line is that anycast amplifies effects of
> bgp/igp twitchiness.

It is always nice when the gut feeling is confirmed from
hard data. Any chance operators [vetted in whatever way you
all choose] could get aggregate data from within their AS?
or AS + downstreams?

Cheers,

joe
--
RSUC / GweepNet / Spunk / FnB / Usenix / SAGE

>> better analysis will be available in a week.
> Never one to grump at free & meaningful info, I didn't beat
> the drum on the 'week' marker. Just wondering if there is a
> horizon to the more detailed/better analysis.

a bit more is at <http://rip.psg.com/~randy/050311.wide-anycast.pdf>

but there will be more at nanog

> Any chance operators [vetted in whatever way you all
> choose] could get aggregate data from within their AS? or
> AS + downstreams?

actually, we are looking for a few good asns to analyse in
some depth. what we want is (for you) to run the probe from
a few places inside the as while recording as much of the
as's internal bgp info as possible in zebra/quagga/mrt
format. we believe we will then be able to tell you a bunch
about your routing.

randy