[nsp] BGP Entries - Comparison

[Robert E. Seastrom]

"Paul Stewart" <pauls at nexicom.net> writes:

> Hi everyone...
> 
> I'm trying to diagnose why our upstream providers have some varying numbers
> of routes coming to us...
> 
> Provider A has 136992
> Provider B has 141471
> Provider C has 121855
> 
> These are three full feeds coming into us... Why the huge difference in
> numbers?  I realize that there will be differences based on connectivity
> etc. but that much??
> 
> I'm currently looking at how many /25 /26 /27 etc. *small* subnets are
> hitting us, as Provider B is sending us thousands of /27 blocks and even
> thousands of /29 blocks....

Provider B is sending you a lot of trash; you can *probably* filter
out longer than /24 with no ill effects whatsoever (modulo
considerations for funky routing tricks with multiple links, multihop
bgp, and intentional deaggregation on your connection to them - if you
have a normal connection it's in all probability OK to filter away).
Provider A is likely sending you something approaching reality
(consensus from my four fairly-decent upstreams is in the
137500-138800 range); provider C is filtering some longer prefixes
from RIR space and old-B-space.

                                        ---Rob