Re: 8600 Diffserv

["Phil Wang" <pywang@nortelnetworks.com>]

Hi Glynn,

I've gone over these 8600 documents and found that 8600 does not provide
a user interface to manage its congestion, sorry. Obviously, 8600 does
the congestion control internally without the user interference. The
document "Reference for the Passport 8000 Series Management Software
Routing Operations (Part No. 207415-B)" gives details about configuring
QoS (Chapter 11).

> Thanks for the reference. However I regret to say I remain unconvinced that
> the 8600 has proper AF functionality. I could find no mention of RED or
> probalistic drop mechanisms anywhere in the document which is consistent
> with everthing else I have been reading.There is simply the assignment of
> the AF DSCP's to particular queues in the WRR scheduler but this does not
> constitute an AF PHB.
These queues are with different forwarding priorities. Ch 11 of that
document states the below (11-1) and provides a secton "Queue
Servicing".

"Within the DiffServ network, the marked packets are placed in a queue
according
to their marking. For example, if a video stream is marked to receive
the highest
priority, it is placed in a high-priority queue. As those packets
traverse the
DiffServ network, the video stream is forwarded before any other
packets."

> 
> I've been looking through a MIB listing in a document Tal sent me but there
> is no obvious mention of anothing to do with RED
No, the MIB does not include it.

> 
> I have also been trying to access the reference manuals referred to in the
> document to see if there is a CLI command for configuring RED. However the
> Baynetworks library URL that Tal sent me has not replied all day. I finally
> found the document on the Nortel website but the attempt to read it
> resulted in another connection timeout. Seems we have some problem with the
>  link to the US - I'll try again on Monday.
I do not find it either.

> 
> So, as I said, I remain unconvinced but I'm willing to be pursuaded.
> 
> 'Aveagoodwegend as we say here (freely translated as 'have a good weekend')
Have a nice weekend too.

Phil
> 
> regards
> 
> Glynn
> 
> At 12:48 PM 9/6/00 -0400, you wrote:
> >Hi Glynn,
> >
> >>
> >> Gentlemen,
> >>
> >> The 8600 documents have arrived and have been very informative. My
> >> understanding of the Diffserv situation is this.
> >>
> >> The Diffserv system requirements document provides a general specification
> >> for Diffserv support in all future devices. Apart from some concern over
> >> the way Weighted Fair Queueing schedulers are described, it all made very
> >> good sense to me and bodes well for the future.
> >>
> >> However the second document specifically on the 8600 Diffserv support makes
> >> it clear that the 8600 does not, at least yet, provide anything like the
> >> level of support described in the first document. In fact Diffserv PHB
> >> support appears to be limited to the Class Selector PHB's although it may
> >> turn out to be possible to implement Expedited Forwarding once the current
> >> effort to finalise the EF definition is completed.  The document confirms
> >> my suspicion that
> >>
> >> 1/ policing is rate based (peak and average) but not token bucket
> >That's true. Actually it uses one rate only.
> >
> >>
> >> 2/ the 8600 does not have a RED like mechanism and therefore cannot support
> >> Assured Forwarding in spite of what is said in one of the public documents
> >> I saw.
> >I do not know what documents you have had, but my documents show that AF
> >should be supported using congestion-avoidance mechanism either drop
> >precedence or RED-alike.
> >
> >>
> >> The first is not a problem for me at the moment but the second has caused
> >> me to revise the proposal I am preparing for Nortel Australia. Again this
> >> is not a major problem but I need to take the drastic step of including a
> >> couple of Cisco routers in the system to provide Assured Forwarding
> >> functionality - I hope this does not cause any heartburn.
> >I still believe that AF forwarding must be provided by 8600. See the
> >attachement.
> >
> >Thanks,
> >Phil
> >
> >>
> >> Thanks for the documents - it was just as well I waited. Now I can proceed
> >> with finalising the draft proposal
> >>
> >>  regards
> >>
> >> Glynn
> >>
> >>