8 Replies Latest reply on Jun 8, 2016 11:39 AM by ophirmaor

    Ethernet Fabric


      Hi all,

      I want to know does Mellanox Ethernet Switches support Trill, SPB or something like these protocols for Ehternet Fabric ?

      does sx1410 Support vxlan ? what about sx1710 ?

      can I pass layer 2 vlan over layer 3 network with Mellanox Ethernet switches ?

      if any docs please reference me


        • Re: Ethernet Fabric

          Hi, we don't support Trill or SPB.

          What do you mean by support of VXLAN? do you mean VXLAN termination? VTEP?

          QinQ is supported on Mellanox switches.


          Which adapters do you have?


            • Re: Ethernet Fabric

              Hi ophir,

              To extent Vlans over Layer3 Network ( I want to mention in spine/leaf Topology for Big Datacenters,) there are two approach, one deploying Trill/SPB and the other is vxlan ( Bridging ), the first question is about Mellanox approach I think VMS, Mlag, MAGP are presented by Mellanox.

              But my question is focused on layer2 over layer 3 with vxlan. Do MLAG/MAGP can cooperate with each other? based on the following pic


              default gateway for server s2 is G1 and S2 is in vlan Red. server S1 is the same but Gateway is G2. these two server wants to be in the same broad cast domain.

              we have to transfer vlan2 over layer 3 network between leaf/spine layer. in this manner vxlan is usefull. so leaf switches must be able to insert vxlan tag.

              does vxlan cooperate with MLAG/MAGP ? do mellanox switches support layer 2 over layer3 ?

              exactly for this scenario what mellanox solution are ? ( any deep technical reference links).

              I hope I could explain what I 'm looking for.

              if there are some ambiguous please tell me to explain more.

              wait for any response.



            • Re: Ethernet Fabric

              Unfortunately, Mellanox can't do any of those things right now. VMS seems to be no more than a bunch of independent switches in a fat tree topology, without any stacking or "chassis virtualization".


              For L2 onto L3 encapsulation, you will need another (pair of) devices to perform those functions, as a couple of routers performing VPLS over GRE tunnels.


              Best regards.