This slide is based several days' worth of statistics
gathered from one of our routers dumping debug ip bgp updates.
Two points need to be made:
1/ it's unfair. It does not look at the flaps
AS 1239 itself is responsible for, although it
does measure all our customer-access ASes. None
of those are in the top 50, although we can still
improve things.
An hour's worth of statistics looking at AS 1239
and others seems to put AS 1239 in 30th place
(about 40 flaps/hour), and matched the longer-term
results quite closely.
Also, we are in trouble when it comes to things
passing through us and sending flaps elsewhere.
This is something we need to work on with our
customers and in our CPE routers.
Alot of the top twenty are Sprint customers.
2/ Flaps are measured by counting up transitions only;
laziness led me to consider a flap as an up-and-down
transition, and I didn't count both of them,
except by eye. The number of flaps, therefore,
is half the number of transitions. The previous
slide is really the number of transitions/second
(ups and downs).
Double one or half the other to suit your tastes.
These results appear to be reproducible elsewhere.
Again, this prints OK on a QMS printer in landscape
mode (qpr -ls).
Sean.
- --
# This is a shell archive. Save it in a file, remove anything before
# this line, and then unpack it by entering "sh file". Note, it may
# create directories; files and directories will be owned by you and
# have default permissions.