6

I heard that once AS receives a BGP update that is longer than 50, it will be dropped. Is this true or just happens with some ISPs specifying that policy? I cannot find any evidence or official documents regarding this. It's great if anyone can give a pointer showing the maximum length of BGP update is accepted.

Cheers,

8

It is possible to filter updates where the AS_PATH length exceeds a maximum value. The command for Cisco is:

bgp maxas-limit

There have been bugs in the past that caused issues when receiving excessively long AS paths, by misconfiguration or exploitation. It is normal to only see 5 or so ASs in an AS path and at most you could see 10-20 if an ISP is doing excessive AS_PATH prepending, so it would not be unreasonable for ISPs to filter at 50 or less.

Recommendation is to filter on AS_PATH length, but doesn’t provide a recommended length.

See (page 58)

https://nsrc.org/workshops/2016/apricot2016/raw-attachment/wiki/Track2BGP/05-BGP-BCP.pdf

Or (page 27)

https://www.ripe.net/participate/meetings/regional-meetings/manama-2006/BGPBCP.pdf

The bug should be patched now, so there is less need, but it does still add to router memory requirements to receive excessively long AS_PATH and maybe the people doing the excessive prepend need to learn the hard way and have their routes dropped.

There is also evidence here that some major ISPs are not filtering

http://seclists.org/nanog/2017/Sep/388

| improve this answer | |
  • 1
    RE “Maybe the people doing the excessive prepend“. If someone is prepending tons of ASN strings and they don’t want the route dropped, perhaps they deserve any education they get. – Mike Pennington Jul 23 '18 at 19:37
  • @MikePennington I agree :) – Karl Billington Jul 23 '18 at 19:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.