Hi, We have been investigating options for optimizing the ovs userspace datapath and hope to post patches for this over the next months. One of the vectors that we have started looking at is the dpcls wildcard classifier. However, in order to understand what our options are to optimize this, we would really like to understand how it gets used in most deployments. A lot of good work was done for OVS 1.11 to introduce megaflows which should reduce the number of flow additions/sec in the datapath. Is there any data that shows, in real world deployments, how often additions need to be made to the fast path considering that the dpcls is now wildcarded?
Thanks, Mark _______________________________________________ discuss mailing list [email protected] http://openvswitch.org/mailman/listinfo/discuss
