L2VPN Meeting - IETF90 (Toronto) 1300-1500 Monday July 21 2014 WG Status Web Page: http://tools.ietf.org/wg/l2vpn/ Chairs: Nabil Bitar         Giles Heron Secretary: Andrew McLachlan Working Group Status and Update Chairs: Giles Heron/Nabil Bitar  10 minutes Status Update Nabil give the status updates. (please see updates) A full agenda at this L2VPN WG in Toronto, mostly centred around EVPN Comments from the room about the microphone being to quiet at the back…problem resolved by audio tech in the room. There have been a few RFCs published since the last working group, 4 of them in particular. We finished e-tree requirements draft, EVPN requirements, VPLS MIB and VPLS inter-domain redundancy. There are a number of drafts at the IESG. these include L2VPN framework which had a few comments back from the AD which the editor needs to take note of. L2VPN EVPN 07 is also at IESG, Adrian had a few comments on that also, which Ali is editing. L2VPN IPLS this has been long in the queue and is moving to historic publication, and we have VPLS MAC optimisation this has been in the process for a few months and we should have that as last version and more to RFC. On PBB EVPN, this has been issued for WG last call, which ended today, hopefully this will move towards RFC now. Now on Trill EVPN draft and short path bridging draft for EVPN. Trill had expired, and the others we will talk about later to see if WG wants to proceed with that. We have a couple of other drafts on VPLS which we had issued WG last call for, but we have very little response to that. Pimp snooping draft we extended comments for 3 weeks and there was zero response, so are people not interested in this draft? We will discuss that later. The VPL broadcast which has expired, do we want to continue this? Draft on MAC flush which has been around for a couple of years, expert review done but no one is picking it up, so also do we want to continue it? VPLS multihoming has very little progression also. Mic John Albert - I think there has been a new VPLS multihoming draft in the last week Nabil - Andrew has been following that, we might have missed that on this slide. On other drafts we have the MPLs TP Mac withdrawal which was in PWE3, this was adopted the home for it was decided to be L2VPN, and an IPR call was done for this. We need to decide VPMS requirements framework draft, this was blocked for some work in PWE3, which was done. However as this draft was based on legacy technology, ATM etc, do we want to continue this now, are there people how want to work on it? VPWS interworking OAM draft which has been around for some time, Giles is going to shepherd that draft now. Therefore I would like to take a poll from the WG. So we will start with the drafts issued for WG last call, pe-tree we have progressed the requirements, the framework is on the way to RFC, so what has happened to the solution draft. Is there any interest in this today? Please raise your hands…..Very few hands. So we will reissue and see where we are on this. Mic Lucy - I thought for WG last call we are not expecting people to vote unless they have comments, so there is no objection to that draft, so why are we to re-issue Nabil - So before it was, just support, support etc which gave no feedback really, however this time only the authors have said they support it. So we will move it to the AD and see what to do, and likely an expert review. No name - I see comments who is a co-author of the draft and he supports it What about PIP-snooping, no authors in the room? Any support, please raise hands. VPMS + VPLS broadcast extensions – no support Ldp opt – Jorge wants to carry it Multihoming – updates needed Thank you for your help with this, now we will move to the presentations. Presentations. draft-sajassi-l2vpn-EVPN-inter-subnet-forwarding-04.txt Ali Sajassi presents the slides Mic Stewart - can we fix the audio? Ali - I will move the mic Stewart - better Nabil - How many people have read the draft? (Many hands) bad audio, no name - how many bits are you going to use? Ali - 32 bits, if it is MPLS then advertisement goes with BGP extended community to indicate what tunnel it is. * - so you indicate if it is a label or a… Ali - in another BGP ex community you indicate tunnel type, and you treat them accordingly. Wim - I’m supporting WG adoption, however you should remove GDP. Ali - I agree we don’t need to for VXLAN, so we can remove it. Nabil - So how many support it (hand), o so we take it to the list. draft-boutros-l2vpn-EVPN-vpws-04 Sami Boutros presents the slides   Mic Lucy - Question on load balancing what kinda of load balancing can you perform based on the MAC Sami - The is no MAC lookup, it’s a point to point service, Lucy - but at the CE side you will need load balancing based on VLAN? Sami - what do you mean load balancing based on VLAN? Lucy - but when you are multi-homed you will have PE1 and PE2 saying they can reach the CP? Sami - the PE1/Pe2 would be advertising the ESI, and advertise it as multi homed. Ali - Basically the VLAN can be reach by both PE1 and PE2 so the remote PE can load balance over this. Ali - for ECMP that is taken care of with entropy label and that is different and flow based. Sami - this is a point to point service. Ali - this intended for SP using EVPN for point to point Nabil - we need to move on. How many people support the draft?(some hands), comments to the list please draft-boutros-l2vpn-vxlan-EVPN-04 Sami Boutros presents the slides Mic Jorge - you don’t support local attachment circuits? If all the traffic is sent with anycast I don’t see how you do it. Sami - I don’t we considered that in the draft. Ali - for locally attached AC we could use a different IP address, and that takes care of it. You are right this is not captured in the draft/ Sami - please send this as a comment on the draft and we can look at it more. Nabil - how many read the draft? (hands) ok good we take it to the list. draft-sd-l2vpn-EVPN-overlay-03.txt Ali Sajassi presents the slides Mic Jorge - A couple of points. First the encoding of the VNDI in the ethernet tag v the MPLS label, this is something that changed from v1 to v2 in the drafts. The other is split horizon. First, so there are implementations already and for some of them they are using the ether tag to encode the global vnid. So the draft is a bit confusing around the mpls label. Ali - you are right someone already mentioned this, and we need to make this consistent. with the Ethernet tag a 0 as per EVPN mode. Jorge - another thing is a mismatch with 20 bit and 24 bit fields in ethernet and BGP. Jorge - on split horizon, based on source IP address right? Ali - yes, source squenching Jorge - I think this is broken for multi-homing and also for some merchant silicon don’t support the checking on source IP, also on underlay like NAT using overlay tunnels where source IP is changed. The check based on source is therefore not enough and we should think about other attributes too. Ali - if you have suggestions then let discuss. Nabil - how many people read the draft? (many hands support adoption. draft-rp-l2vpn-EVPN-usage-02 Jorge Rabadan presents the slides. Mic no questions Nabil - how many read the draft? Ok same as previous, we can take it to the list. draft-rabadan-l2vpn-EVPN-prefix-advertisement-02 Jorge Rabadan presents the slides Mic Robin (Huawei) - for IP prefix we need to co-ordinate with L3VPN? Nabil - This was discussed with the L3VPN WG chair Wim - Yes this was discussed and we will have to present and we needed a new route type as there is no MAC next hop and no ESI next hop in IPVPN. The use cases in the draft outline this. Ali - yes we have been in discussion with L3VPN chairs, the draft talks about the differences. Martin (poss, bad audio) - this has been discussed in L3VPN, then they can come and express opinion there. Nabil - Sounds like we need to hold a discussion in L3VPN. draft-li-l2vpn-EVPN-pe-ce-01 Shunwan Zhuang presents the slides Mic no questions Nabil - how many have read this? A few hands, so we need those some feedback on this please. draft-singh-l2vpn-bgp-VPLS-control-flags-01 Ravi Singh presents the slides Mic no questions Nabil - ok we will take i to the list. draft-keyupate-l2vpn-fat-pw-bgp-01 Sami Boutros presents the slides Mic no questions Nabil - how many have read this? Ok very few hands, so we need those some feedback on this please. draft-fang-l3vpn-virtual-pe-05.txt Luyuan Fang Mic Lucy - in this model do you assume this vPE will be part of the IGP component? Luyuan - you can do both or neither. no name - I don’t see anything for QoS for vPE as compared to hardware model for QoS Luyuan - It is not the same, hardware and software behaviour differently. no name - do you have a protocol between the vPE and the control and data plane on other devices? Luyuan - If you are using a controller then you construct your FIB via the controller. no name - but I belief you need a control plane protocol between the devices Luyuan - you use the controller of any other protocol you are using. Nabil - there has been some dialogue here and perhaps we need some clarification. Luyuan - Yes perhaps we need some clarification. draft-hao-l2vpn-inter-nvo3-EVPN-00 Weiguo Hao presents the slides Mic Ali - I wonder if the VNID you consider here is it a global or local. Weiguo - it is local Ali - and in the DC network? The VNID is for a given VNE, is it local, or is it the same VNID for a given broadcast domain? Weiguo - *audio bad* Ali - I think this is covered in EVPN overlay inter-AS, have you had a chance to look at it? Also if it is locally significant, then you can do it with an MPLS label between MPLS and IP. If it is Global then EVPN talks about using a gateway function, draft DCI-EVPN, Jorge’s draft. Weiguo -This is a heterogeneous type Ali - yes this is exactly what we are taking about, and they are described in the drafts I mentioned. Weiguo - I think we should talk about it offline. draft-rabadan-l2vpn-EVPN-optimized-ir-00 Jorge Rabadan presents the slides Audio too bad to understand that this point to understand Meetings ends 2 mins after audio break up.