Carlos Alberto Kamienski wrote:
>
> > >3.2, list, DSCP.
> > > Why is it only a single DSCP ? Also, is there only a single Flow ID
> > > per SLS, or can there be many ?
> >
> > A set of DSCPs is definitly needed.
>
> I think the concept of service should be enforced rather than to imply it
> by some DSCP(s). What I mean is that, it may not be possible to implement
> correctly a service just by observing the DSCP, traffic conformance
> parameters and performance guarantees.
>
> For example, it is possible to use a particular PHB (like EF PHB) to
> deploy different services, but if you want to deploy an end to
> end service by concatenating some intra-domain services like the Virtual
> Wire PDB, you must follow this PDB specification. This has an implication
> even in the network topology.
>
> The deployment of such a service may require a previous engineering in
> the network. Then, the negotiation process (a protocol ?) should be based
> on services that networks are really able to implement.
> This leads to something like the QBone BB idea of globally
> well-known services (GWS). In this case a GWS ID should be included in the
> SLS along with other parameters that specify paticular levels of this
> services.
>
> Comments ?
>
Hi Carlos,
you are right in the sense that a service should not particularly map on
a DSCP. In fact, the choice of DSCP may be signalled back from the
network to the customer, upon request of a set of certain transport
characteristics for a certain flow (in the SLS). As an extreme example,
given enough overdimensioning in a domain, the virtual leased line may
be offered to best effort traffic...
As far as globally well-known services concerns, it might help, but i am
not sure if it is really necessary.
regards,
Yves
> Regards,
> Carlos
>
> ----------------------------------------------
> Carlos Alberto Kamienski - cak@cin.ufpe.br
> Federal University of Pernambuco - Brazil
This archive was generated by hypermail 2b29 : Mon Oct 16 2000 - 09:01:37 CEST