Re: SONET/Ethernet clock tolerance
Bruce,
You are only partially right about some of what you say. Mostly there is
still some perception and paradigm issues. The "SONET Lite" PHY that has
been proposed is WAN compatible, not WAN only. It is not full SONET and
does not have the standard SONET clock tolerance issues.
The clock tolerances for regenerators (LRE) are very relaxed from those of
line terminating equipment (LTE). It is the LTE that does the multiplexing
of the smaller TDM payloads into a full OC192 payload and puts the full
SONET overhead on it, scrambles it, and sends it out. The regenerator does
a line clock recovery, unscrambles the SONET frame, uses the section
overhead information to determine performance and fault issues, rescrambles
the SONET frame, and sends it out. The regenerator re-times the signal
using the recovered clock, not a Stratum clock. In the process the
regenerator removes any signal bit jitter that has been introduced by the
various forms of dispersion that occur in optical fiber over any distance.
Because it does not do the multiplexing and does not have the clock
tolerance issues that the LTE has, a regenerator is a lot less expensive
than an LTE.
There is not a major desire to preserve the investment in the 10Gb SONET
equipment in the field that will be used to support 10GbE. Primarily, most
of the build out to support 10GbE will be installing new equipment, the
latest, and least expensive forms of regenerators and amplifiers. The real
reason to preserve the SONET and SDH framing is to leverage equipment and
technology that is already in the field and available from vendors. No new
technology will have to be developed and deployed to support 10GbE over long
haul systems. That makes building out and supporting 10GbE over long haul
systems a lot less expensive than it would otherwise be. This is the
ultimate in leveraging existing technology.
Thank you,
Roy Bynum
----- Original Message -----
From: Bruce Tolley <btolley@xxxxxxxxx>
To: Osamu ISHIDA <ishida@xxxxxxxxxxxxxxxxxxx>; <rtaborek@xxxxxxxxxxx>
Cc: <stds-802-3-hssg@xxxxxxxx>; <devendra.tripathi@xxxxxxxxx>
Sent: Tuesday, March 28, 2000 11:14 PM
Subject: Re: SONET/Ethernet clock tolerance
>
> Ishida-san:
>
> At 11:06 AM 3/29/00 +0900, Osamu ISHIDA wrote:
> >Here 10 Gigabit Ethernet will require only ONE Path STS-192c in the
payload.
> >I don't believe re-using expensive SONET regenerator for 10GbE is a good
> >choice
> >unless you have already invested in it.
>
> Exactly, I think you hit the nail on the head. I have listened very hard
> for months now to the SONET discussion and the debates on the LAN vs WAN
PHY.
>
> I have always interpreted the requirement for the so-called WAN PHY as an
> installed base issue. Some WAN folks believe there is a requirement to
> preserve the investment in installed SONET-TDM equipment which is driven
by
> a return on invested capital business model.
>
> There are other models and other ways of building networks. It is
> refreshing to hear other points of view.
>
> Bruce Tolley
> Cisco Systems