Re: QSPEC: <Excess Treatment>
Gerald Ash <[email protected]>
| Newsgroups | gmane.ietf.nsis |
|---|---|
| Message-ID | <[email protected]> |
Hannes, > Hi all, > > " > The default excess treatment in case that none is specified is that > there are no guarantees to excess traffic, i.e. a QNE can do whatever > it finds suitable. > " > > I think this sentence should make it clearer that "default" in this case > does not refer to a specific value of the <Excess Treatment> but rather > to the absence of the parameter. OK. > " > When excess treatment is set to 'shape', it is expected that the > QoS Desired object carries a TMOD parameter. Excess traffic > is to be shaped to this TMOD. When the shaping causes > unbounded queue growth at the shaper traffic can be dropped. If > excess treatment is set to 'shape' and no TMOD parameter is given, > the E flag is set for the parameter and the reservation fails. > " > > When a QoS Desired Object contains 2 TMODs, one for the regular QoS > treatment and another one for the shaping of excess traffic then how do > you differentiate the two? The text does not say anything about this > case. This should be specified by the applicable QOSM. There are many possible instances of interacting QSPEC parameters where the QOSM will need to specify the intended behavior. > > When excess treatment is set to 'remark', the excess treatment > parameter MUST carry the remark value, and the remark values and > procedures MUST be specified in the QOSM specification document. For > example, packets may be remarked to drop or remarked to pertain to a > particular QoS class (DSCP value). In the latter case, remarking > relates to a DiffServ model where packets arrive marked as belonging > to a certain QoS class/DSCP, and when they are identified as excess, > they should then be remarked to a different QoS Class (DSCP value) > indicated in the 'Remark Value', as follows: > > Remark Value (6 bits): indicates either drop (set to 0) or DSCP > value to remark packets to when identified as > excess > > The value of 0 is unnecessary since this can already be accomplished by > value 0 for the Excess Trtmnt. > Is the encoding of RFC 2474 used? Yes. > > Ciao > Hannes > Jerry _______________________________________________ nsis mailing list [email protected] https://www.ietf.org/mailman/listinfo/nsis