On 5/23/23 16:16, Robin Jarry wrote: > Aaron Conole, May 23, 2023 at 15:32: >> I think one issue I have with this is that the name is a bit >> misleading. Control plane, from OVS perspective, would be like OpenFlow >> communications. This is more like a traffic steering mechanism. >> >> Maybe it would help to call it something like "traffic-based-rps" or >> something like that (but not clear what would be best). It is really a >> way to steer specific traffic to a distinct rxq. >> >> WDYT? > > Actually, "packet-steering" was one of the first ideas I had for this > feature, but I thought it may be confusing. I am weary of reusing the > "rps" acronym as it may be confused with a linux specific feature: > > https://docs.kernel.org/networking/scaling.html#rps-receive-packet-steering > > The feature introduced in this patch is relying on RTE flow which has > nothing to do with Linux. However, I agree that the name should reflect > that we are steering traffic into a dedicated rxq. How about these > ideas? > > options:isolated-rxq=lacp,... > options:rxq-isolate=lacp,... > options:rxq-steering=lacp,... > > I personally prefer "rxq-isolate".
'rxq-isolate' will be confused with 'other_config:pmd-rxq-isolate'. Same likely goes for the 'isolated-rxq'. 'rxq-steernig' may be confused with 'other_config:tx-steering'. But this can be argued that it's essentially similar functionality, so should be named similarly. Maybe we can double down on that and use options:rx-steernig=rss|rss+lacp|... with 'rss' being a default configuration? Best regards, Ilya Maximets. _______________________________________________ dev mailing list d...@openvswitch.org https://mail.openvswitch.org/mailman/listinfo/ovs-dev
