On Thu, Dec 18, 2014 at 11:59 AM, Benson Schliesser <[email protected]> wrote: > > Behcet Sarikaya wrote: >> >> >> Quoting from my previous message: "one could imagine the NVE >> imposing an underlay DSCP in IP2, e.g. to discriminate between >> tenants." >> >> No it is not the issue of coping DSCP from IP1 to IP2 especially in tenant >> based QoS. >> How can a DC operator apply QoS on each tenant separately when IP1 is >> created by a VM that does not know the VNI? > > > The DSCP in the underlay IP header can be copied, imposed, translated, etc.
It seems you did not follow up recent email exchanges. We are talking about tenant based QoS. The tenant id or VNI is known by NVE. So NVE has to do the QoS assignment. > The NVE can decide what to put into those DSCP bits.Thus to reiterate my > previous assertion, which David Black recently quoted: there is no QoS gap > that needs to be addressed in the overlay encap layer. As we said in Rev 02, if tenant based QoS is not used then yes DSCP bits can be copied from VM's values. For tenant based QoS, Ethernet frame QoS bits are used, as we explained in Rev 02. We also defined a mapping to DSCP bits from Ethernet frame QoS bits after the encapsulation is removed at the receiving NVE. > > By the way, I have looked at the diff between -01 and -02 of this draft and > I still see no explanation of any problem. Please read Rev 02. We can make a revision to explain tenant based QoS better if you like. > On the other hand, I do see an > arbitrary redefinition of existing QoS mechanisms. As it stands, this is not > something that NVO3 will pursue. Please see above. > > Unless you can explain a problem that isn't addressed by the existing > mechanisms there is no reason to continue discussing it. Two issues we are addressing: -tenant based QoS -QoS framework in the overlay network What more can we do for nvo3 :-) Happy Holidays. Behcet > > Thanks, > -Benson > _______________________________________________ nvo3 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nvo3
