Matt-

Per discussion during IDR/SIDR meeting Friday, there may need to be some
text in the security considerations around the attack vector of sending
many updates with long (but valid) AS_Paths, since the analysis Sriram
provided indicated a correlation between the length of the AS Path to be
validated and CPU impact or convergence time impact. Sounds like we don't
have a problem with long paths comprised of  multiple prepends because
Pcount means that there aren't multiple signatures, but validly signed
strings of unique ASNs could still trigger this problem, whether they are
generated accidentally or maliciously. We may also need to make the
observation that rapid route churn with these sorts of prefixes are likely
to make the impact even higher, such that it may be advisable to implement
route flap dampening to provide additional protection against route churn.


That said, I'm not sure that this is strictly a security considerations
addition, because we may need to make some suggestions in the design to
protect against these sorts of scaling problems.
I'm not sure if we can use analysis of existing routing table data to
identify an 80/20 threshold of the longest credible AS Path and put a
recommendation on a max_as_length value for the limit for protection
against this attack, or whether we cannot recommend a value, and have to
simply acknowledge that this is a problem and recommend that implementers
provide a knob for setting a limit on AS Path such that implementations
can intercept and drop long AS_Paths before the BGPSec machinery tries to
validate them. Perhaps both, i.e. A reasonable default limit that can be
raised or lowered if a user sees fit. Alternatively, we could suggest a
fall-back method whereby excessively long-path BGPSec updates are
deprioritized or throttled to avoid causing adverse performance impacts,
perhaps in conjunction with existing RFD implementations for routes that
are both long-path and churning.

I can try to contribute some text once the WG weighs in on what direction
they want to take to address this, but I wanted to at least get the
discussion started.

Thanks,

Wes George



On 10/27/14, 7:40 PM, "Matthew Lepinski" <[email protected]> wrote:

>Just posted the -10 revision of the document.
>
>The only normative change was to decouple BGPsec validation from
>Origin Validation. (This is a normative change to the validation
>algorithm in Section 5.) This is based on working group discussions at
>IETF 90 and confirmed on the list in September.
>
>Additionally, there were some text changes made, including a reference
>to the AS-Migration document (and an accompanying text change
>regarding pCount=0).
>
>I have a couple of improvements to the security considerations text
>that I wasn't able to fold into this version of the document.
>
>Other than improving the security considerations, I am not aware of
>any open issues in this particular document.
>
>- Matt Lepinski
>
>_______________________________________________
>sidr mailing list
>[email protected]
>https://www.ietf.org/mailman/listinfo/sidr


This E-mail and any of its attachments may contain Time Warner Cable 
proprietary information, which is privileged, confidential, or subject to 
copyright belonging to Time Warner Cable. This E-mail is intended solely for 
the use of the individual or entity to which it is addressed. If you are not 
the intended recipient of this E-mail, you are hereby notified that any 
dissemination, distribution, copying, or action taken in relation to the 
contents of and attachments to this E-mail is strictly prohibited and may be 
unlawful. If you have received this E-mail in error, please notify the sender 
immediately and permanently delete the original and any copy of this E-mail and 
any printout.
_______________________________________________
sidr mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/sidr

Reply via email to