Hi Ali, > -----Original Message----- > From: Ali Sajassi (sajassi) [mailto:[email protected]] > Sent: Thursday, November 05, 2015 11:16 AM > To: Jeffrey (Zhaohui) Zhang <[email protected]>; Samer Salam (ssalam) > <[email protected]>; Dennis Cai (dcai) <[email protected]> > Cc: [email protected] > Subject: Re: Questions on draft-sajassi-bess-evpn-l3vpn-multihoming > > > Hi Jeffrey, > > > On 11/5/15, 7:19 AM, "Jeffrey (Zhaohui) Zhang" <[email protected]> wrote: > > >Hi, > > > >I have a few questions: > > > > On the S-PEs, the pseudowires from the Access PEs are terminated onto > > VRFs, such that all pseudowires within a given redundancy set > > terminate on a single IP endpoint on the S-PEs. To achieve this, the > > S-PEs in a given Redundancy Group are configured with the same > > Anycast IP and MAC addresses on the virtual (sub)interface > > corresponding to the VRF termination point. > > > >Is the virtual interface (like) an IRB interface? > > No IRB interface in this case. This is the virtual interface/sub-interface > address. > > > > > Since the S-PEs are running in EVPN single-active redundancy mode, > > the S-PEs would advertise an Ethernet AD route per vES with the > > single-active flag set per [RFC7432]. Since only the DF S-PE has its > > access pseudowire in Active state, only that device would establish > > an eBGP session with the CE and receive control and data traffic. > > > >Will the IRB/virtual interface on a non-DF PE be up? I assume it is - > >even though the PW is not active. Would the non-DF PE keeps trying to > >establish the eBGP session with the CE? Would that cause issue to the > >session between CE and the DF PE? > > The backup PW will be in stand-by mode (not up) and thus there won¹t be a > eBGP session established between non-DF and the CE. Only when the non-DF > becomes DF, then eBGP session gets established.
Whether it is an IRB interface or not, is the interface considered up (at layer 3) when the PW is in the stand-by mode? I understand that the EBGP session won't be established, but will the standby PE try to establish it? Would that attempt cause trouble to the established session (assuming the CE will get messages from the standby PE via the primary PE)? > > > > > The > > DF S-PE advertises host prefixes that it receives, from the CE over > > the eBGP session, to other PEs in the EVI using EVPN route type-5, > > with the proper ESI set. Remote PEs learn the host prefixes and > > associate them with the ESI, using the advertising PE as the next-hop > > for forwarding. > > > >Would he DF S-PE advertise other prefixes received on the eBGP session? I > >assume so but the text only says host prefixes. > > That¹s Correct. There will be other prefixes. We¹ll take care of it in the > next rev. > > > > > Other S-PEs in the same Redundancy Group as the advertising PE will > > receive the same EVPN route type-5 advertisement, and will recognize > > the associated ESI as a locally attached vES. > > > >What is the RT that limits the routes to be imported only the PEs in the > >same redundancy group? Is it that all PEs in the same EVPN instance will > >import the routes? Or is it that different redundancy groups will have > >different EVPN instances? I assume it's latter (since we need one IRB > >interface for each redundancy group?). > > The RT will be the RT associated with the vES/ES. Will the vES/ES belong to a single EVPN instance only? > > > > > The withdrawal of the Ethernet > > Segment route serves as an indication to the backup S-PE to go active > > (i.e. act as a backup DF), and activate its pseudowires to the Access > > PE. The withdrawal of the Ethernet A-D route triggers a "mass > > withdraw" on the remote PEs: these PEs adjust their next-hop > > associated with the prefixes that were originally advertised by the > > failed PE to point to the "backup path" per [RFC7432]. > > > >Can you elaborate the procedure for the PEs to adjust their next-hop? > > Please refer to sections 8.4 and 14.1.1. Part of my confusion came from the (wrong impression) that EVPN procedure is also used between the APE and SPE (for vpws). Now that it is not the case, would the same concept in this draft apply even if the CE is connected via regular ACs? Thanks. Jeffrey > > > > >I don't see reference to draft-ietf-bess-evpn-vpws. Is that used at all? > > Evpn-vpws is not used but it can be used to reduce the PW provisioning to > only single-side. > > Cheers, > Ali > > > > >Thanks. > >Jeffrey _______________________________________________ BESS mailing list [email protected] https://www.ietf.org/mailman/listinfo/bess
