Skip to main content

Minutes IETF102: pce
minutes-102-pce-00

Meeting Minutes Path Computation Element (pce) WG
Title Minutes IETF102: pce
State Active
Other versions plain text
Last updated 2018-07-20

minutes-102-pce-00
PCE Working Group Meeting – 13:30-15:00 Wednesday Afternoon session I

   o Chairs: Jonathan Hardwick, Julien Meuric (remote)
   o Secretery: Dhruv Dhody
   o Slide - https://datatracker.ietf.org/meeting/102/session/pce
   o Etherpad - http://etherpad.tools.ietf.org:9000/p/notes-ietf-102-pce?useMonospaceFont=true
   o Meetecho:  http://www.meetecho.com/ietf102/pce  

1. Introduction
1.1. Administrivia, Agenda Bashing (chairs, 5 min)
1.2. WG Status (chairs, 20 min) [25/90]

Jonathan Hardwick: For draft-ietf-pce-lsp-setup-type, 2 authors have not 
                   replied to AUTH48 . Please respond! 

Daniel King: For draft-ietf-pce-inter-area-as-applicability, Is there still 
             interest in the WG to update this document? 
Jon: Significant "editorial" comments to improve the documents and the major 
     issue and there was no information on the the stateful PCE
Daniel King: I can take the lead and delegate some of the work to the 
             co-authors but want to be sure if there is interest to update 
             the document. 
Poll: Is there interest in inter-area-as applicability further. 
      A few raised hands. 
Deborah Brungard: This is a milestone on PCE charter since 2013. 

Young Lee: For draft-ietf-pce-wson-rwa-ext, shepherd review done by Danielle, 
           Young to address the comments next. 
Julien Meuric (Meetecho): The proto write-up for RWA is even ready to go.

Jon: Request for document shepherds for the WG documents in the queue. Please 
     help! 
    
Dhruv Dhody: For draft-ietf-pce-pcep-yang, already updated with examples, 
             idnits, request for the yang doctor review;
Dhruv: The other documents are moving well,  don't have any pending action 
       items, please review and send comments to the list and we hope to move 
       them quickly. 
Haomian Zheng: For draft-ietf-pce-enhanced-errors, aim to make the draft ready 
               by next meeting.
Adrian Farrel: For draft-ietf-pce-pcep-flowspec, moving along well, pushing 
               implementation side, a question - to what extend this needs to 
               harmonize with SR-TE policy. Opinions? 
Jon: This is an open question from last meeting as well. We need to handle 
     this as the draft progress. 

2. WG I-Ds
2.1. Update on Association Drafts (Dhruv Dhody, 10 min) [35/90]
Various Drafts

No questions


2.2. Stateful H-PCE and ACTN (Young Lee, 15 min) [50/90]
draft-ietf-pce-hierarchy-extensions-05
draft-ietf-pce-stateful-hpce-05
draft-ietf-pce-applicability-actn-06

No questions

3. PCE as central controller
3.1. PCECC Drafts (Dhruv Dhody, 15 min) [65/90]
draft-zhao-pce-pcep-extension-for-pce-controller-08
draft-zhao-pce-pcep-extension-pce-controller-sr-03

Igor Bryskin: Are PCInitiates done in sequence or parallel?  
Dhruv: Parallel
Igor: How does the controller know? 
Dhruv: We usually start from the egress to ingress, but we have PCRpt message 
       to clearly tell if the download was okay and re-compute path if not! 
Andrew Stone: If PSLP-ID (LSP) is mandatory, how does this apply to SR?  
Dhruv: Coming back to this later in the SR mode
Igor: what do you "make" during the Make-Before-Break? Entire path? 
Dhruv: Yes
Igor: It may be just a path revision, rather than full E2E Make-Before-Break 
      as done in RSVP-TE.
Dhruv: We dont differentiate between changes needed only in a few nodes v/s 
       entire path, Will consider the terminology changes with you off-line 
       (if needed). 
Andrew: Can you put the correlation between PCEP session and router id in an 
        individual draft? As it might have applicability beyond this! 
Dhruv: Can be done! 
Haomian: For re-delegation, more than PCE down, it is also worth to consider 
         the overloading on any PCE, and balance some nodes/LSP to another 
         PCE. 
Dhruv : This is consistent with RFC8231 overload mechanism and we don't 
        differentiate between LSP state and this! 
Igor: In case of network re-convergence, can you program multiple updates to a 
      node on a single message?  
Dhruv: Yes, this is allowed via existing PCInitiate message.
Jon: Allow for capability advertisement for PCECC-SR only (without basic PCECC 
     mode). 
Dhruv: This can be added, the current mechanism could be a historical because 
       of the older version where we had a "new message" defined in the basic 
       mode! 
Jon: Do you want to adopt both PCE-CC and SR? 
Dhruv: Yes
Poll:  On draft-zhao-pce-pcep-extension-for-pce-controller; 
   How many have read the draft? ~around 12
   How many would like to see WG adoption? ~ around 12;
Poll: On draft-zhao-pce-pcep-extension-pce-controller-sr; 
   How many have read the draft? ~around 8
   How many would like to see WG adoption? ~ around 8;
Jon: There is good support in the room so we will ask the same questions on 
     the mailing list.
   
3.2. New Extentions (Cheng Li, 15 min) [80/90]
draft-li-pce-controlled-id-space-00
draft-li-pce-sr-path-segment-00
draft-li-pce-sr-bidir-path-00

Jeff Tentsura: You are not signaling co-routed in the bi-directional draft?  
               Any reason not include co-routed bi-directional co-routed? 
Cheng Li: It may be co-routed in this association, we did not restrict or 
          specify in the draft, may do that later. 
Jeff: For accounting, path ID needs to be informed on all nodes and PCEP 
      session may be needed to all nodes, like flooding! 
Cheng Li: Path-ID is informed to ingress and egress and not inspected by the 
          transit for basic use case. 
Jeff: The beauty of SR is we removed state from the network and this is also 
      the ugly part. 
Dhruv: In the current version, we inform path ID to Ingress and Egress only, 
       for the path accounting use-case this might change but now with PCECC SR 
       we are moving towards PCEP as SBI and thus we need to handle this 
       upfront anyways. 

4. Previously Discussed Topics
4.1. ACTN VN Association (Daniele Ceccarelli, 10 min) [90/90]
draft-leedhody-pce-vn-association-05

Igor: This is a good idea but could it be done in a generic way rather at a 
      "VN association" specifically. Just like a group of LSPs in a mesh.
Daniele Ceccarelli: Association is anyway generic, This is a new association 
                    type, you could define another! 
Igor: Think of a TE tunnel set. 
Daniele: You can define another association.   
Young: What Igor is asking is can it cover more than VN, with a generic TE 
       tunnel set included. 
Poll: Who has read the draft ~ around 12
      a good based to adopt? ~ around 10
Jon: Discuss with Igor on list and I will discuss with my co-chair and take 
     it to the list! 
      
-- Meeting Done, See ya in Bangkok --