Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: [802.3ae] Wan Interface Sublayer




Hi, Mike,

I believe your general observations are correct. In addition, the network diagram
you have drawn is also valid. The WIS and SONET networks are in principle
separate, as far as the WIS clause of 802.3ae is concerned.

However, the conclusion that you have drawn - i.e., that the WIS costs more
in parts, money, power, efficiency, etc., and is hence unlikely to be used in any
application other than an interface to an ELTE - is not universally held. There
has been a great deal of discussion and argument on the reflector on this subject
over the past two years. For example, one of the dissenting views I have heard is
that the benefits of dealing with a single interface type might outweigh the incremental
costs of using a WIS even where one is not necessary. I would expect that, in the
end, the choice would be made by the enterprise user or service provider, based on
their policies, goals and network models.

Paul Bottorff presented a fairly detailed justification for the WIS in a presentation
at the March 2000 IEEE 802.3ae meeting; you can find it at:

http://grouper.ieee.org/groups/802/3/ae/public/mar00/bottorff_2_0300.pdf

Best regards,

- Tom Alexander
WIS Scribe

-----Original Message-----
From: Ayers, Mike [mailto:Mike_Ayers@bmc.com]
Sent: Friday, August 24, 2001 2:49 PM
To: stds-802-3-hssg@ieee.org
Subject: RE: [802.3ae] Wan Interface Sublayer


> From: Tom Alexander [mailto:Tom_Alexander@pmc-sierra.com]
> Sent: Friday, August 24, 2001 11:00 AM

> I echo James' response. The objective for the WIS was Path
> transparency
> (not Section and Line transparency). Therefore, the WIS SPE
> data rate is
> compatible, the clock tolerance meets SONET Minimum Clock
> requirements,
> and the WIS SPE is formatted properly, with the necessary
> Path Overhead
> functionality. This greatly simplifies the so-called ELTE
> (see T1X1.5/2001-095),
> reducing it to a straightforward Path relay function. There
> was no intent to
> allow a 10GBASE-W interface to be connected optically to an OC-192c
> interface.

        Gotcha.  Enlightenment is imminent.  A few more thoughts and
questions.

        It seems to me that the need for an ELTE means that WIS and SONET
networks must remain separated.  This being the case, I don't see a general
use for WIS.  In other words, I see WIS as useful to do this:

                        _____
                   ____/     \____
                __/               \
               /        SONET      \_
         +---> |        cloud        \ <---+
         |      \_____           ____/     |
         |            \_________/          |
      +--+---+                          +--+---+
      | ELTE |                          | ELTE |
      +--+---+                          +--+---+
         |                                 |
  +------+---------+              +--------+-------+
  |      W         |              |        W       |
  |                |              |                |
  | 802.3ae switch |              | 802.3ae switch |
  |                |              |                |
  | R     R     R  |              | R     R     R  |
  +-+-----+-----+--+              +-+-----+-----+--+
    |     |     |                   |     |     |
    |     |     |                   |     |     |
  +-+-+ +-+-+ +-+-+               +-+-+ +-+-+ +-+-+
  | A | | B | | C |               | D | | E | | F |
  +---+ +---+ +-+-+               +-+-+ +-+-+ +-+-+

        But I do not see any other reason for WIS usage, since it requires
more parts (and therefore cost, and power) and has a slower data rate than
its R counterparts.  Am I missing something here?


/|/|ike