Document: draft-ietf-pce-sr-p2mp-policy
Title: PCEP extensions for SR P2MP Policy
Reviewer: Yingzhen Qu
Review result: Has Issues

Hi,
I have been selected as the Operational Directorate (opsdir) reviewer for
thisInternet-Draft.

Document: https://datatracker.ietf.org/doc/draft-ietf-pce-sr-p2mp-policy/13/
Reviewer: Yingzhen Qu
Review Date: 2025-12-05
Intended Status: Standards Track

---
Summary: Has Issues

Major Issues:
The draft doesn't have a section for "Operation Considerations." Section 9 is
"Manageabiity Considerations", but with a couple of sub-sections as "TBD". What
the document proposed is quite complex. There are many moving parts, such as
TE, PTIs, PCE/PCC driving updates, so some operation consideration about state
convergence, configurations etc would be helpful.

Please include some scalability consideration, especially in terms of number of
multicast groups and the sizes of multicast groups.

Please add some clarifications on what happens when a partial branch fails, how
PCE/PCC detect failures, and how traffic is rerouted etc.

The "IANA Considerations" section should match the description text in the
previous sections.

The document needs an editorial pass. For example:
There are Candidate Path/Candidate path/candidate path in the document. Please
keep it consistent. Multiple places are missing a period mark "." at the end of
a paragraph.

Some detailed comments below:
177      engineering criteria and any additional Service Leave Agreements
178      (SLAs) that is used to construct the tree.
s/Service Leave Agreements/Service Level Agreements

248      similar concept as draft [draft-ietf-pce-segment-routing-policy-cp].
The reference needs to be updated to RFC 9862. There are multiple instances in
the document.

265      PCE MAY also calculate and download additional information for the
266      replication segments, such as protections next-hops for link
267      protection.
s/protections next-hops/alternative next-hops?

289      optimization of a CP with in a SR P2MP policy. Each CP can have
s/with in/within

291      There are used for Make Before Break (MBB) and global optimization
292      procedures.
This sentence doesn't parse.

315      It should be noted that the [draft-hb-spring-sr-p2mp-policy-yang] can
316      provide further details of the high level P2MP Policy Model.
I'd suggest removing referece to this old and expired document.

416      With varying encoding rules for the SR-P2MP-LSP- IDENTIFIER TLV which
s/SR-P2MP-LSP- IDENTIFIER TLV/SR-P2MP-LSPID-TLV

422      identify a cross connect. A cross connect is a incoming SID and set
s/a incoming/an incoming

424      CCI objects contains the incoming SID and the outgoing interfaces
s/contains/contain

470      * PCE then sends a PCUpdate to the Root node indicating the
471       association information (CP) and implicitly binds the CP to the
472       installed CCI information.

501      - PCE will then send a PCUpdate to the root indicating the
502       association information (Candidate path) , and implicitly
503       indicate it to bind to the latest CCI information downloaded.
These two steps are identical. Should the descripton also be the same?

513     Transit, Bud and Leaf nodes in the SR P2MP Tree using a PcInitiate
s/PcInitiate/PCInitiate. There are multiple instances in the document, such as
line 529 and 533.

516      * - PLSP-ID: value MUST be set to zero and will be assigned by PCC.
For PCC-Init procedure, Root sends PLSP-ID to PCE, why MUST it be set to zero
in PCInitiate message?

559      PCUpd message. Association object MUST be present forCP PCUpdate
s/forCP/for CP

594      To remove a single CP, PCE sends PC Initiate message
s/PC Initiate/a PCInitiate

Section 4.3.3.4
Q: If there are multiple CPs in one SR P2MP policy, does the sequence of
removing each CP matter when removing the policy?

Section 4.3.4
Q: what if the CP already has to PTIs? shall the 2nd PTI (not active) be
removed first?

628      The transit and leaf nodes SHOULD be able to accept traffic from both
629      PTIs to minimize the traffic outage by the Make Before Break process.
Q: If there is a new node added to the network, it's possible that the existing
active PTI forms a loop. Is this considered?

665      interface is down. There can be 2 method to protect the primary
666      interface.
s/2 methond/two methods

Section 4.3.
6For the two protection methods in Figure 1, please consider adding a
comparison of advantages/disadvantages of each method.

713      The SR P2MP Policy and its Replication segment can be delete by the
714      PCC or by the PCE. to delete the SR P2MP Policy all the CP associated
s/delete/deleted
s/to/To

730      For PCC to delete a CP, Root send a PCRpt message with the R bit of
731      the LSP object set and all the fields of the SR-P2MP- LSP-ID TLV set
s/send/sends
s/SR-P2MP- LSP-ID/SR-P2MP-LSPID-TLV

739      For PCC to delete a PTI, Root send a PCRpt message with the R bit of
740      the LSP object set and all the fields of the SR-P2MP- LSP-ID TLV set
Same as above.

750     4.3.8.2. PCE Initiated
579     4.3.3.4. Conveying active state and cleanup
These two sections are partially duplicates.

756     4.3.9. Fragmentation
Is this section needed? it doesn't seems to define anything new. However the
document should add some operation considerations when there are fragmentations
and how to ensure fragmentations are handled consistently.

996      SHOULD BE ignored if present.
s/SHOULD BE/SHOULD be

1155  (Add/Remove/Replace-all), the PcRpt and PcInit messages are extended
s/PcRpt and PcInit/PCRpt and PCInitiate

1227    Instance-ID to identify a PTI with in the CP.
s/with in/within

Section 5.7.2
Q: https://www.iana.org/assignments/pcep/pcep.xhtml#pcep-objects has an early
allocation of object-type value 3. Am I missing something here?

Thanks,
Yingzhen


_______________________________________________
Pce mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to