Re: [EFM] RE: Pause frame usage in transport networks
Siamack,
Only if it does not violate any company confidential information issues can
specific technology be discussed. In some cases, where success and failure
anecdotal information can sometimes help with technical decisions. This is
the reasons that I, after being away from WorldCom for a period of time,
felt comfortable with discussing the details of some of the research work
that done concerning IP over DWDM and Ethernet over SONET.
Thank you,
Roy Bynum
At 02:12 PM 3/5/2003 -0500, Siamack Ayandeh wrote:
>Roy,
>
>Now this is somewhat different from the way the previous statement came out.
>Mapping data to a fixed size TDM pipe may cause latency variation as a
>function of
>how much data is buffered. A small component in order of several
>microseconds is
>due to SOENT/SDH 125us frame structure.
>
>Regards, Siamack
>ps I am not sure if we are supposed to talk company specific stuff in the
>WG or its
>mailing list
>
>Roy Bynum wrote:
>
> > Siamack,
> >
> > The 125us that I was referring to was worst case under close to 100%
> > utilization. Not all frames had that much latency variance, even at that
> > loading. Some of the frames in the traffic stream had a latency variance
> > that was closer to the bit time delta of the largest to smallest frames,
> > which is closer to the inherent worst case latency variance of native 802.3
> > Ethernet. Some frames had very little, if any latency variance.
> >
> > Under best case conditions, data traffic streams can have almost no latency
> > variance, particularly when switching constant size frames. This is true
> > of almost all Ethernet only data switches. I am not sure if it is true of
> > IP/MPLS switches, but it could be under certain conditions.
> >
> > I would interested in hearing more about the what your company is doing.
> >
> > Thank you,
> > Roy Bynum
> >
> > At 09:36 AM 3/5/2003 -0500, Siamack Ayandeh wrote:
> >
> > >Roy Bynum wrote:
> > >
> > > > Regardless of the protocol that is used, all TDM transmission
> facilities
> > > > induce a 125us latency variance at the customer data link
> layer. This is
> > > > because the transmission convergence mapping is based on a specific
> payload
> > > > window that occurs every 125us in SONET framing.
> > >
> > >Roy,
> > >
> > >I don't think this statement is quite correct as not the entire 125us
> > >frame needs
> > >to be mapped/switched at once. Current generation SONET/SDH
> framing/switching
> > >equipment have a latency variance that is an order of magnitude less
> than the
> > >125us frame interval.
> > >
> > >Regards, Siamack