[Pce] Re: Opsdir early review of draft-ietf-pce-stateful-pce-vendor-04

[[e-mail]]

Hi Samuel,

Thanks for the new version and the clear responses.
Please see inline.


Original


From: SamuelSidor(ssidor) <[e-mail]>
To: 肖敏10093570;
Cc: [e-mail] <[e-mail]>;[e-mail] <[e-mail]>;[e-mail] <[e-mail]>;
Date: 2024年08月12日 15:41
Subject: RE: Opsdir early review of draft-ietf-pce-stateful-pce-vendor-04




Hi Xiao,
 
Sorry for delay, most of issues/comments should be fixed in attached version. Please see responses for remaining issues/comments:
 
Your comment:
" Section 3, my first feeling is that this section should list all Stateful PCEP objects in which the Vendor Information TLV may be contained, however after checking Section 3 of RFC 7470, I found it says "Further specifications are needed to define the position and meaning of the Vendor Information TLV for specific PCEP objects". Then I think this section should either define the Vendor Information TLV for each Stateful PCEP object or state something like what's said in Section 3 of RFC 7470"
 
Response:
This draft is only about carrying the <VENDOR-INFORMATION> object in stateful PCEP messages, it does not make any changes to VENDOR-INFORMATION-TLV processing, where RFC 7470 continues to apply. The draft is even explicitly inheriting all rules from Section 3 of RFC 7470 (see: "All the procedures are as per section 3 of [RFC7470].")
[XM]>>> OK, that's fine.
 
Your comment:
" Section 4.6, compared to what's said in Section 6.6 of RFC 7470, it seems what's said here is a little bit too simple."
 
Response:
Based on Section 4 of this draft, all manageability requirements and considerations are inherited from RFC7470 (and from other RFCs), including that longer text from Section 6.6. Are you missing any specific recommendation?
[XM]>>> In Section 6.6 of RFC 7470, it seems that the provided mitigation method covers only the direction from PCC to PCE, then how about from PCE to PCC?
 
Your comment:
"Considering that multiple Vendor Information Objects/TLVs of multiple LSPs can be carried in the Stateful PCEP messages, it can be imagined that in some cases the amount of Vendor Information would become too huge to be processed by the receiver timely. In other words, some kind of congestion may happen due to the added Vendor Information. So it's helpful to the reader/implementer if some mitigation method can be provided here."
 
Response:
There is no change against RFC7470, where multiple requests with Vendor Information object could be also included in single PCReq message. Individual vendors are not exposing structure of their Vendor Information Objects, so it is not expected (or it will really rare) that some specific implementation will include Vendor Information Object from different vendor. Section 6.1 of RFC7470 (which is inherited by this draft) is also proposing that inclusion of vendor specific information may be configurable, so it can be disabled if not really needed (e.g. if vendor of PCC and PCE are not same and including vendor specific information is useless). There was also recent discussion about potentially optimizing encoding of Vendor Information object, but conclusion seems to be that if any optimization needs to be done, then it should be generic optimization applicable to other PCEP objects, which is out of scope of this draft.
[XM]>>> Thank you for the detailed explanation. If appropriate, some text on potential out-of-scope optimization can be added here, it's up to you. :-)

Cheers,
Xiao Min
 
Thanks a lot,
Samuel
 

From: [e-mail] <[e-mail]> 
 Sent: Monday, August 12, 2024 8:51 AM
 To: Samuel Sidor (ssidor) <[e-mail]>
 Cc: [e-mail]; [e-mail]; [e-mail]
 Subject: Re: Opsdir early review of draft-ietf-pce-stateful-pce-vendor-04


 
OK, no problem. :-)
Looking forward to your new version and more discussion if needed.
Cheers,
Xiao Min

Original

From: SamuelSidor(ssidor) <[e-mail]>



To: 肖敏10093570;



Cc: [e-mail] <[e-mail]>;[e-mail] <[e-mail]>;[e-mail] <[e-mail]>;



Date: 2024年08月07日 20:40



Subject: RE: Opsdir early review of draft-ietf-pce-stateful-pce-vendor-04




Thanks a lot Xiao for review and comments.
 
 We are discussing changes required to the draft with co-authors. We will get back to you soon.
 
 Regards,
 Samuel
 
 -----Original Message-----
 From: Xiao Min via Datatracker <[e-mail]>  
 Sent: Tuesday, August 6, 2024 9:42 AM
 To: [e-mail]
 Cc: [e-mail]; [e-mail]
 Subject: Opsdir early review of draft-ietf-pce-stateful-pce-vendor-04
 
 Reviewer: Xiao Min
 Review result: Has Issues
 
 Summary: I've reviewed this document and I believe this document is on the right track. I have no major concern but several minor ones. Besides, there are a number of nits and ungrammatical sentences, I'm also not good at this, so just to name a few.
 
 Major issues: None.
 
 Minor issues: As below.
 Section 2, it says "Different instances of the object can have different Enterprise Numbers". I believe a normative language is more suitable than *can*, MUST or MAY? It's supposed to be MAY. Section 3, my first feeling is that this section should list all Stateful PCEP objects in which the Vendor Information TLV may be contained, however after checking Section 3 of RFC 7470, I found it says "Further specifications are needed to define the position and meaning of the Vendor Information TLV for specific PCEP objects". Then I think this section should either define the Vendor Information TLV for each Stateful PCEP object or state something like what's said in Section 3 of RFC 7470.
 Section 4.2, it says "Any standard YANG module will not include details of vendor-specific information", and then it provides  a suggestion on how the standard YANG module MAY be extended. I assume the mentioned extension applies only to a proprietary YANG module, if that's the case, then I don't see much value to mention the extension. Section 4.6, compared to what's said in Section
 6.6 of RFC 7470, it seems what's said here is a little bit too simple.
 Considering that multiple Vendor Information Objects/TLVs of multiple LSPs can be carried in the Stateful PCEP messages, it can be imagined that in some cases the amount of Vendor Information would become too huge to be processed by the receiver timely. In other words, some kind of congestion may happen due to the added Vendor Information. So it's helpful to the reader/implementer if some mitigation method can be provided here.
 
 Nits/editorial comments: As below.
 Abstract Section, s/may then be/may be then.
 Section 1, s/(LSP-DB)/(LSP-DB)); s/added new messages in PCEP/add new messages to PCEP; s/[RFC7470] defined/[RFC7470] defines; s/It also defined/It also defines; s/to also include/to include. Section 2, s/be used on a single PCRpt message/be contained in a single PCRpt message. Section 3, SRP needs expansion in first use; s/All the procedures as per/All the procedures are as per; s/defines the Enterprise Numbers are allocated by IANA/defines the Enterprise Numbers allocated by IANA; s/clarifies that the IANA registry described is/clarifies that what the IANA registry describes is. Section 4.4, s/Verify Correct Operations/Verifying Correct Operations. Section 7, s/PCEP also support/PCEP also supports.