|
Sue – I will not reinforce your inappropriate insertion of the discussion of possible RFC 7752 deficiencies into a thread which is reviewing draft-ietf-idr-te-pm-bgp.
This is wrong – please stop it. You want to discuss issues with RFC 7752 – please start a separate thread – directed at the WG (of course) – and state your position as a WG member as to what
you feel needs to be done. Then others can comment and the WG can decide whether they agree work needs to be done. Also - and for the third time - draft-ietf-idr-te-pm-bgp is NOT SR related – please also stop inserting a discussion of SR/BGP-LS into the review of this draft. (Apologies for the harsh tone – but it is becoming increasingly frustrating to watch this thread being misdirected.) Les From: Susan Hares <shares@xxxxxxxx>
Les: I will respond to this information point by point after I have responded to Ketan who sent me messages regarding his draft prior to yours. However, I would
like your feedback on whether you believe RFC7752 has security that is equivalent to, less than, or greater than a trusted domain?
The spring routing architecture (RFC8402) indicates that be default SR operates within a trusted domain. It says further:
“Traffic MUST be filtered at the domain boundaries. The use of best practices
to reduce the risk of tampering within the trusted domain is important. Such practices are discussed in [RFC4381] and are applicable
to both SR-MPLS and SRv6.” If you believe that RFC7752 describes a trusted domain [per RFC4381],
RFC7762 does not state that it is a trusted domain.
If you believe the RFC7752 has stronger security
than a trusted domain, can you let me know
why you think it is stronger than a trusted domain.
If you believe that RFC7752 describes security which
is less than a trusted domain, please let me how it is less.
The RFC7752 text is below for ease of the IDR and secdir readers.
Thank you for providing this feedback,
Sue ------------ RFC7752 security section (section 8.) 8. Security Considerations Procedures and protocol extensions defined in this document do not affect the BGP security model. See the Security Considerations section of [RFC4271] for a discussion of BGP security. Also refer to [RFC4272] and [RFC6952] for analysis of security issues for BGP In the context of the BGP peerings associated with this document, a BGP speaker MUST NOT accept updates from a consumer peer. That is, a participating BGP speaker should be aware of the nature of its relationships for link-state relationships and should protect itself from peers sending updates that either represent erroneous information feedback loops or are false input. Such protection can be achieved by manual configuration of consumer peers at the BGP speaker. An operator SHOULD employ a mechanism to protect a BGP speaker against DDoS attacks from consumers. The principal attack a consumer may apply is to attempt to start multiple sessions either sequentially or simultaneously. Protection can be applied by imposing rate limits. Additionally, it may be considered that the export of link-state and TE information as described in this document constitutes a risk to confidentiality of mission-critical or commercially sensitive information about the network. BGP peerings are not automatic and require configuration; thus, it is the responsibility of the network operator to ensure that only trusted consumers are configured to receive such information. From: Les Ginsberg (ginsberg) [mailto:ginsberg@xxxxxxxxx]
Sue – Inline. From: Susan Hares <shares@xxxxxxxx>
Les: I apologize if my email message was unclear. We both agree that your draft is not related to SR routing. SR routing is related to BGP-LS as a transport mechanism
for information. I agree that RFC7752 had traffic engineering information. However, that traffic engineering information almost got that draft rejected by the IESG at the time.
As my previous message to this list indicated, we got agreement on RFC7752 based on limiting that information and the assurance that BGP-LS nodes were deployed on a separate set of nodes. Expanding the traffic engineering information beyond RFC7752 re-opens
all the security issues and questions from RFC7752’s original review. [Les:] So far, you are the only person who seems aware of this. I am not saying you are wrong – I am just saying my private attempts to get more context
for this have thus far failed – and you have not provided any documentation of this. If your statement is accurate (again – not saying it isn’t) – it also seems most unfortunate (and I am being “kind” here) that this was not mentioned in
the course of the four years that draft-ietf-idr-te-pm-bgp has taken to progress to this point. The security directorate reviewer is asking these security questions. The security directorate does have people with both routing and security experts. SR routing is also expanding the information past the original RFC7752. The expansions requested by SR routing also re-open those original security questions
and issues. [Les:] I do not know why you mention SR here since we both agree this draft is not SR related. One way to answer these questions is to provide a RFC7752bis with an updated security section. If you agree with this approach, I suggest simply referring to
a RFC7752bis that in your security section. If you disagree that an update to the RFC7752bis is required, we can start a thread on that point.
[Les:] There is no RFC7752bis draft.
J So you are asking me to reference a non-existent document? I understand that you (at least) would like to have one – which is perfectly legitimate – though you should go through the normal WG process to take this
work on – correct? But this dodges the question as to whether draft-ietf-idr-te-pm-bgp has a dependency on enhanced security. So far, you are the only person making this claim
– and several folks (including myself) have expressed a different POV. I think you at least have to provide a justification for this dependency before we introduce it and get some support for your position – since this will mean draft-ietf-idr-te-pm-bgp would
be stuck in MISSREF state until this currently non-existent draft becomes an RFC. Les Did this message clarify my earlier brief message? Do you want to continue to discuss the need for RFC7752bis?
Cheerily, Sue
From: Idr [mailto:idr-bounces@xxxxxxxx]
On Behalf Of Les Ginsberg (ginsberg) Sue – One of us is confused.
J draft-ietf-idr-te-pm-bgp is not related to Segment Routing. Those words do not appear anywhere in the document. Nor is there a reference to any SR document. Further, RFC 7752 includes traffic engineering information (see
https://tools.ietf.org/html/rfc7752#section-3.3.2 ) so the suggestion that we are introducing a new attack vector by defining some additional(sic) TE codepoints does not make sense to me. I appreciate that there are other drafts on your mind which are SR related – but this is not one of them. Could you please update your response with these points in mind? Thanx. Les From: Susan Hares <shares@xxxxxxxx>
Robert, Benjamin, and Yoav:
I agree these context of these issues are not specific to this draft. However, traffic engineering information does provide information which is a tempting attack
vector. The original RFC7752 described a different purpose with restricted usage that SR routing extensions do not adhere to in BGP. Since Spring WG shows that operators
are interested in the extended use, it may be time to examine the RFC7752bis or other solutions that takes care of these security issues.
My job as a shepherd is to point out these issues per draft for the IESG and security directorate in order to obtain the correct feedback. As a WG chair, I have
pointed out these issues, but the WG has these drafts on WG LC without the extra security. Without RFC7752bis with additional security in the base document, I am working as a shepherd make the manageability and security sections as clear as possible.
If the feedback from the security directorate review or the IESG is that we need to obtain a solution for RFC7752bis that describes and handles these security
issues, I will be glad to support fast-tracking this issue within the WG. If an offline discussion with Benjamin, Yoav, the IDR chairs, and Alvaro would speed this along, I can set this up early next week. It would be helpful to have
this offline discussion before the IDR interim session on 10/26. Thank you for all your comments.
Sue
From: Idr [mailto:idr-bounces@xxxxxxxx]
On Behalf Of Robert Raszuk Hello Benjamin, Not sure if you have spotted similar comment made to IDR regarding this topic, but your comment seems to indicate that here we are about to define ways to carry nicely scoped IGP information into BGP. Well that has already happened with
RFC7752 and your comment or for that matter Yoav's remarks are indeed spot on but to the security discussion on RFC7752 and IMO not any follow up extensions of it. Sure - as observed by Sue - one may argue that providing more information about the network to the potential attacker makes the network weaker, but the cure for that is to prevent the leaks and reduce probability of intercepting new information
by unauthorized parties. BGP-LS is already defined in a new SAFI what by itself does provide nice level of isolation. RFC7752 is pretty clear on that too and says: "BGP peerings are not automatic and require configuration; thus, it is the responsibility of the network operator to ensure that only trusted consumers are configured to receive such information." If someone would be still concerned about configuration mistakes and negotiating SAFI 71 or 72 to those who should not get this data I recommend we reissue the RFC7752 as -bis version and restrict the scope of the distribution even further
by mandating default use of NO-EXPORT community with ability to overwrite it for the selective eBGP peers. Or perhaps we could progress Jim's One Administrative Domain draft (draft-uttaro-idr-oad-01). In either case while both of your comments are great they seems a bit late in the game here or at least targeting wrong document. Kind regards, Robert. On Fri, Oct 19, 2018 at 2:27 AM Benjamin Kaduk <kaduk@xxxxxxx> wrote:
|
