Sean Hefty wrote:
From what I understand while reading your proposal, is that it is quite different then what what suggested in the original RFC. I don't think it makes sense to implement the host side of this before there's agreement on the over-all solution namely how the host side design/code plugs to the management scheme at the SM side.
I don't believe that my proposal and the SA side proposal are incompatible. We should be able to design the host side stack somewhat independent from a specific SA implementation. It needs to be to support alternative SA implementations.
Say I am with you on the argument that the host side implementation need not be dependent on a specific SA implementation. However, for validation purposes it should be testable with some SA...
But even with that, the host side architecture implemented by your patches to IPoIB and the RDMA CM is --different-- from the one proposed in the RFC.
Specifically the point I see we need to have a debate is your approach of --not-- using the Service ID to derive QoS (ie FL/TC and SL)
I find it more constructive to have a discussion based on a short host side architecture description that covers at least QoS for IPoIB and RDMA CM based ULPs, rather then based on series of patches.
Or. _______________________________________________ general mailing list general@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
