RE: [Diffserv] Model - queues - replacement text

[Andrew Smith <andrew@xxxxxxxxxxxxxxxxxxx>]

OK, so we need suitable RECOMMENDATION text instead, at all places in the
model where re-ordering might be introduced (it's not something specific to
the "queue" section, is it?).

Are we saying that my proposed text:

"this does not preclude implementions which support operations on the FIFO
data structure that remove or examine packets e.g. for use by discarders.
The only constraint on such operations is that they not add packets back
into the FIFO in a different order."

is too restrictive? Can someone provide a suitably realistic example of why
this is too strict? If so then I think we need a less misleading name for
this thing than "FIFO".

Andrew


> -----Original Message-----
> From: Brian E Carpenter [mailto:brian@hursley.ibm.com]
> Sent: Wednesday, December 15, 1999 1:18 PM
> To: Andrew Smith
> Cc: diffserv@ietf.org
> Subject: Re: [Diffserv] Model - queues - replacement text
> 
> 
> Well, never mind the ASICs then. RFC 2474 says
> 
> >    It is RECOMMENDED that PHB implementations do
> >    not introduce any packet re-ordering within a microflow. 
> 
> The text in the model must be consistent with this.
> 
>   Brian
> 
> Andrew Smith wrote:
> > 
> > That's fine for specific PHBs but we are talking here about 
> the generic
> > model of a queueing thing for all current PHBs and as many 
> future PHBs as
> > possible. Not sure I understand the relevance of 
> MF-classifier based ASICs
> > to this.
> > 
> > Andrew
> > 
> > > -----Original Message-----
> > > From: Brian E Carpenter [mailto:brian@hursley.ibm.com]
> > > Sent: Wednesday, December 15, 1999 7:46 AM
> > > To: Andrew Smith
> > > Cc: 'Dan Grossman'; diffserv@ietf.org
> > > Subject: Re: [Diffserv] Model - queues - replacement text
> > >
> > >
> > > Andrew Smith wrote:
> > > ...
> > > > > structure that remove or examine packets (e.g., for use by
> > > > > discarders), as
> > > > > long as no operation reorder packets belonging to the
> > > same microflow.
> > > >
> > > > [Andrew] Suggest we not get into re-ordering issues here to
> > > any granularity
> > > > less than "per-queue": I think it suffices to say here
> > > "this does not
> > > > preclude implementions which support operations on the FIFO
> > > data structure
> > > > that remove or examine packets e.g. for use by 
> discarders. The only
> > > > constraint on such operations is that they not add packets
> > > back into the
> > > > FIFO in a different order."
> > >
> > > Well, some months back we had implementors *insisting* on the
> > > per-microflow
> > > language rather than per-queue, so that the diffserv model would
> > > be compatible with MF-classifier based ASICs. I think 
> that was for the
> > > specific case of AF, but I suspect the argument is generic.
> > >
> > >    Brian
> > >
> > 
> > _______________________________________________
> > diffserv mailing list
> > diffserv@ietf.org
> > http://www.ietf.org/mailman/listinfo/diffserv
> > Archive: http://www-nrg.ee.lbl.gov/diff-serv-arch/
> 

_______________________________________________
diffserv mailing list
diffserv@ietf.org
http://www.ietf.org/mailman/listinfo/diffserv
Archive: http://www-nrg.ee.lbl.gov/diff-serv-arch/