Hi!

> I have read the document (draft-ietf-opsec-bgp-security-01) and I’m in 
> support of publication this document as a BCP. Some comments/suggestions 
> below.  I have not had chance to go through all the comments you have 
> received so far on this list. Hence, sorry if some of mine seem like a repeat 
> of comments from others.  

No problem! I haven't been able to answer to all the great coments we've 
received so far. Thanks for your support that's appreciated!

> At NIST, we had put together BGP security recommendations document several 
> years ago. You may consider referencing it:
> 
> R. Kuhn, K. Sriram, and D. Mongomery, “Border Gateway Protocol Security,” 
> NIST  Special Publication 800-54. 
> http://csrc.nist.gov/publications/nistpubs/800-54/SP800-54.pdf

I wasn't aware of it sorry. I believe this is highlighting that having an IETF 
BCP would be useful as no network operator can be aware of all the great 
material avalaible everywhere.

> Some nits/comments follow:
> 
> Consider defining inbound/outbound upfront.
> 
> p.3, sec 1, para 1: ACL is access control list

OK

> p.3, sec 1, para 1: this quoted phrase needs further clarity and can be 
> rephrased / explained:  “…. to avoid filtering transit traffic if not needed”

Clarified

> section 5.1.1.1 title change: IPv4 special purpose prefixes

OK

> section 5.1.1.2 title change: IPv6 special purpose prefixes

OK

> section 5.1.2: replace “…described in this section if it is not capable of…” 
> with “…described in this section if they are not capable of…” 

OK

> sec 5.1.2.1, para 1: replace “…keep checking prefixes in the IANA allocated 
> address space …” with “…keep checking prefixes in the IANA allocated IPv4 
> address space …”

OK

> sec 5.1.2.1, para 1: replace “…IPv4 prefixes they receive have been 
> allocated…”  with  “…IPv4 prefixes they receive in BGP updates have been 
> allocated……”

OK

> sec 5.1.2.2, para 1: replace “…To overcome this risk…”  with  “…To partially 
> mitigate this risk……”

OK

> sec 5.1.2.3, para 3: replace “…a script can build for…”  with  “…a script can 
> be built for……”

I don't think so. Full sentence is below so we are talking about the fact that 
the script builds something:
"With these two mechanisms a script can build for a given peer the list of 
allowed prefixes and the AS number from which they should be originated"

> sec 5.1.2.3, para 4: comment about “…only use information from sources 
> representing the five current RIRs…”  Question: Why not include the mirrored 
> IRR data of major ISPs as well? I understand such data are readily available 
> in the RADB.

OK, changed to:

One could check a popular IRR containing many sources (such as RADB, the 
Routing Assets Database) and only select as sources some desired RIRs and 
trusted major ISPs

> sec 5.1.2.3, para 5: replace “…may quickly vary over time…”  with  “…may 
> frequently vary over time ……”

OK

> sec 5.1.2.3, para 5: replace “…could even been considered…”  with  “…could 
> even be considered ……”

OK

> sec 5.1.2.4, last para: The validation procedure described is inaccurate and 
> incomplete. So (per Randy’s suggestion as well) you may simply refer to RFC 
> 6811 here. It may also be worth including RFC 6907 (the SIDR RPKI use cases 
> document) as a reference here.

Yes we'll work further on the section globally.

> Sec 5.1.4: It would be helpful if a definition of “local AS” is provided here.

OK. Same here we should clarify few things here.

> Sec 5.1.5.2: Acronyms pMTUd and uRPF should be spelt out.

OK, will appear in accronym&definition section of -02 

> Sec 5.2.3.1, para 1: replace “is the same than the one for” with “is the same 
> as the one for”  

OK

> Section 8, last bullet: Is there a practical example when an exception 
> (accepting your own AS number in AS-path) is required?   

Well I've personnaly seen that when customers have a single AS for several 
sites and are interconnected through various ISPs. We are usually using other 
mecahnisms like as-override but I'd assume some would prefer to change default 
BGP bahaviour and agree on the risks. I am not sure we wanted to put an example 
in the document.

Again, thanks a lot for your participation!

Jerome

_______________________________________________
OPSEC mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/opsec

Reply via email to