global BGP instability

Would anyone care to give some operational insight into the surge
of BGP routing instability this morning between appx 08:00 and 10:00
GMT? Turned on and off like a lightswitch, as seen from GRADUS
( -- we saw cleanly correlated and sustained
increases in the prefix announcement and withdrawal rates reported by
all GRADUS feeds during that window. --jim

p.s. By the way, we haven't seen the kind of sustained increase in
"maintenance cycle" routing instability that we were expecting after
the SNMP news broke. Some increase in the background noise, but nothing
major overall. Good job, guys :slight_smile:

You can track the proceeding at

Comments can be filed in the proceeding through the
FCC's Electronic Comment Filing System

The comment period closes 75 days after publication in
the Federal Register - and it has not been published
yet so the clock has not started counting.

More information about the Computer Inquiries
themselves can be found at including a
guide descibing what the rules are today


Several providers announced scheduled maintenance last week.
UUNET/Worldcom listed 60+ different cities in its maintenance
announcement for Tuesday 2/19, and more on Thursday. Abovenet
said they would be upgrading some of their routers on Monday 2/18
and Tuesday 2/19. Sprint had a few maintenance items scheduled
for Tuesday. It wouldn't surprise me if other providers had
also scheduled maintenance for Tuesday morning.

There doesn't seem to much variation in the total BGP announcements,
average latency or packet losses, so I don't think it was someone
tickling lots of BGP routers.