> On Apr 29, 2015, at 6:34 AM, Russell Bryant <[email protected]> wrote: > > I wonder how realistic it would be to have 65k ports and hit this as a > limit. If your deployment has many containers in each VM, it seems like > we could hit that much more quickly than traditionally seen before.
I had the same thought around containers. I think we need to assume we'll hit 65k ports in the whole system pretty quickly. > If a protocol can fit a whole UUID in tunneled packets, I suppose we > could optionally switch to using the logical port UUID later? The whole UUID seems like a lot of overhead. I'll start a new thread to discuss how we want to handle end-point identification. If we did a schema like use a 24-bit logical network id and a 16-bit logical port number that should scale fine. Even if we do a good job with broadcast suppression, 65k ports in a subnet is still probably pushing practical limits. I'll get that new thread started. --Justin _______________________________________________ dev mailing list [email protected] http://openvswitch.org/mailman/listinfo/dev
