Re: Is the diffServ preemptive ?

Geunhyung Kim wrote:
> 
> Hello all,
> 
> I have basic question about diffServ architecture.
> As far as I know, AF flow may use the excess bandwidth when there is underload, end the minimum bandwidth should be guaranteed even though
> the link state is overloaded.

But that is not a question about the diffserv architecture. The 
diffserv architecture (RFC 2475) applies equally to all PHBs, not
just AF, and says nothing about guarantees or bandwidth.

> 
> Is that correct ?

Assuming your question applies to AF, i.e. RFC 2597, AF does not define
an AF "flow". It defines a set of AF behaviors allowing the creation
of a set of AF traffic aggregates. If I assume your question applies to
an AF traffic class (i.e. all the data packets carrying the code
points {AFn1,AFn2,AFn3} for a given n), then the relevant part of the RFC is:

>    A DS node MUST allocate a configurable, minimum amount of forwarding
>    resources (buffer space and bandwidth) to each implemented AF class.
>    Each class SHOULD be serviced in a manner to achieve the configured
>    service rate (bandwidth) over both small and large time scales.
> 
>    An AF class MAY also be configurable to receive more forwarding
>    resources than the minimum when excess resources are available either
>    from other AF classes or from other PHB groups.  This memo does not
>    specify how the excess resources should be allocated, but
>    implementations MUST specify what algorithms are actually supported
>    and how they can be parameterized.
> 
So, the answer to your first question is more subtle than "yes" or "no".
 
> If it is correct, when the higher priority flow enters into network, the excee bandwidth that the lower priority flow used may be preempted by the higher priority.
> Even in this case, the higher priority flow can not use the bandwidth that the lower priority flow required minimally.
> 
> Is that correct ?

Neither diffserv, nor AF, defines "priority" among traffic classes.
Let's be very clear about that: there is no priority ordering between
AF1x, AF2x, AF3x, AF4x. They each have their assigned resources as
defined in the above extract.

   Brian Carpenter
 
> 
> Thanks in advance,
> 
> Geunhyung
> 
> None of us is as smart as all of us
> ==========================================
> Geunhyung Kim
> 
> E-mail: [email protected]
> 
> Tel: +82-54-279-5655
> Fax: +82-54-279-5699
> 
> Networking & Distributed Systems Lab.
> CSE
> POSTECH
> ===========================================

_______________________________________________
diffserv mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/diffserv
Archive: http://www.ietf.org/mail-archive/working-groups/diffserv/current/maillist.html