[ofa-general] QoS RFC / how its linked to the networking stack QoS?

[Or Gerlitz]

Sean Hefty wrote:
> Giving this a little more thought, IPoIB uses a single path to each IP 
> destination.  You're asking about per socket ToS, which requires the 
> ability to handle multiple paths to each remote IP address.  Both the 
> original RFC and my proposal for the host stack limit all IPoIB traffic 
> to a single QoS level.

> Is there a real need to expand IPoIB ToS capabilities?

I guess that at the long run, maybe yes.

As this point an architecture that derives the QoS for IPoIB based on 
the partition this IPoIB device lives in make sense.

This means that the first implementation might ignore skb->priority and 
the users would be instructed to use different partitioning for 
different QoS levels.

Again, this has to be clearly stated in text before becoming a code (in 
this case, before becoming a "non code" b/c you are going to not 
implement the linkage..)

Or.