IETF 87 - Routing Area WG  
     Chairs: Alvaro Retana and Alia Atlas
     Scribe: Acee Lindem (acee.lindem@ericsson.com)        

 *  Document Status (Alvaro Retana) - See Slides
     - 2 New RFCs 
     - Composite Links Requirement Draft Last Call Coming 
     - Alia Atlas - Wants to finish Composite Link (CL) by the end of year. 

 * Muticast FRR Tree Notifications (Ice Wijnands) - See Slides. 
     Alia: Explain why this approach is interesting for multicast versus
           unicast due to longer acceptable convergence times. 
     Alvaro: How many read? A couple. 

  * An Architecture for Multicast Protection Using MRTs (Ice Wijnands) -
     See Slides.  
     Lucy Young: Will support single failure or multiple? 
     Ice: The mechanism will support a node or link failure. 
     Lucy: Multiple failures?
     Ice: No - Like other IPFRR mechanisms, multiple failure
          require reconvergence. 
     Rob Shakir - How much does it diverge from optimal flooding path? 
     Alia: MRT Draft includes analysis. 
     Rob: Tree will be sub-optimal. 
     Ice: This trade-off is similar to unicast. 
     Rob: Will affect IP TV. 
     Ice: Other solutions can handle this. Can talk offline. 
     Alvaro: Who has read? 6 or so people. 

   * MRTs Drafts: Architecture, Extensions, Algorithms, IGP 
     Extensions (Alia) - See slides. 
     - Architecture Draft:
       Stewart Bryant: Is this really ready to go strait to Proposed
        Standard status rather than experimental?
       Alia: I have strong faith in the algorithm. This is for
             alternatives. 
       Alvaro: We will evaluate status based on WG consensus
               and implementations.  
       Stephane Litkowski: What is the cost of algorithm? 
       Alia: Chris will show analysis. It depends on how good the 
             alternatives are. 
       Stewart: Multiple MRT algorithms?
       Alia: Profile is where algorithm parameters are set for an
             MRT Island. 
       Stewart: Is domain wide consistency required?
       Alia: MRT islands formed per profile. 
       Stewart: Always consistency between nodes in MRT island?
       Alia: Yes
       Robin Li: Wants standards track rather experimental. 
     - Algorithm Draft
        Alia: Wants WG adoption. 
        Alvaro: Need people to read it. 
        Stephane: How many vendors are implementing prototype?
        Alia: At least two.
     - Analysis of MRT topologies (Chris Bowers) 
       Acee: MRT computation metric must match primary computation?
       Alia: No - MRT is independent of primary path. 
       Stewart: What would happen if you didn't use primary metrics?
       Alia: Has ideas on analysis but may not have time to analyse 
             this variation.  
     - OSPF Extensions for MRT (Alia Atlas)
     - LDP Extensions for MRT  (Alia Atlas)
     
   * Applicability of LDP Multi-Topology for Unicast Fast-reroute
     Using Maximally Redundant Trees (Robin Li) - See Slides. 
     Stewart: Is this the reconvergence loop?
     Robin: No - it is a different loop. 
     Alia: Multiple processes will choose different hops for same
           prefix.
     Stewart: Can't fix problems in base topology with 
              IPFRR mechanism. (This was in respect to one 
              particular LDP use case that was viewed a 
              misconfiguration by some). 
     Alia: Mobile Backhaul use case is more interesting. 

   * Node protecting Remote LFA (Stephane Litkowski) - See Slides. 
     Chris: Did you look at slides to reduce PQs? 
     Stephane - Yes, have discussed polices for reduction. 
     Stewart: Did you look at the LAN case? 
     Stephane: No
     Stewart: Would you look at Naming's draft for this purpose?
     Stephane: Yes 
     Alia: How many have read? Not too many. 

   * rLFA Computation (Chris Bowers) - See slides. 
     Stewart: Can't you have valid Q nodes that won't be found 
       in forward SPF?
     Chris: No
     Stewart: Still worried about assumption that forward SPF 
        and reverse SPF will not strictly visit same nodes. 
     Alia: Do rSPF on PQ node selected. 
     Chris: Hypothetical formula. 
     Rob Shakir: Can do the same with Stephane's algorithm if
       check for node protection is made along the way. 
     Alia: Path may not go through next-next-hop. 
     Chris: Could do rSPF from next-next-hop. 
     Rob: Next-next-hops are more constrained. 
     Stephane: Likes this approach since he previously proposed 
      it. Where to apply PQ selection policies is not obvious
      and requires computation itself.
     Chris: WG adoption?
     Alia: rLFA Node Protection base inclusion? Merge two
       drafts?  
     Stewart: Base rLFA is ready with link protection. No 
              need to further describe the algorithm or delay
              for node protection maturation. 
     Alia: Need rLFA algorithm description for analysis and 
           operator confidence in technology. This requested 
           in draft comments. 
     Alvaro: Need to discuss offline as well as draft merge. 
     Alia: Need to look at computation time frame of rLFA
           versus rLFA node protection. 

   * Problem Statement for MTU Configuration (Vic Liu) - See slides.
     George Swallow: How will this relate to the STATUS BOF work?
     Adrian Farrell: Speaking as AD. This WG is a good place to 
                     start - bounce off Operations WG as well. 
     Alia: Good discussion on mailing list. Let's continue.  
     Vic: Would welcome other authors and contributors.