On Mon, Mar 2, 2015 at 4:39 AM, Thomas Graf <tg...@noironetworks.com> wrote: > On 02/20/15 at 06:23pm, Jesse Gross wrote: >> VXLAN vs. Geneve are pretty much the same for the basic feature set >> including hardware offload and ECMP. Obviously Geneve gives you more >> space and choice for future extensibility. It's not quite ubiquitous >> yet but the next release of Ubuntu will have kernel support out of the >> box and expanded OVS userspace support should hopefully be fleshed out >> more shortly, so I think all of that is OK for the OVN timeline. >> >> But you already knew what I was going to say, right? :) > > Not opposed to Geneve but I'd argue that Geneve is still lacking a bit > offload availability wise compared to VXLAN. It seems like this should > be a operator policy decision anyway.
Coming back to this after processing my vacation email... I think in many cases, Geneve is actually similar enough to VXLAN that if you only use the base header then NICs that were designed to offload VXLAN can offload Geneve as well without changes. And if you need to use more than the base VXLAN header then the choice of encapsulation is not really a question of performance but of whether you can implement the given functionality at all. In any case, I agree that this is just about the default and it makes sense to allow operator choice where possible. The biggest direct impact of this decision would be assumptions about how much space is available in the header. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
