[EFM] RE: clarification on compliance definition

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

Let me make one thing clear - time to market is not the ultimate motive of standards.

(It may be important to you, but the world does not revolve around AllOptic.)

The IEEE 802.x family of standards establishes a framework that encompasses many

different data networking systems. If we follow the model that you are recommending,

then all the 802.x standards should be merged into 802.3. Why not? It would be faster,

easier, less organizational overhead, leverage the Ethernet 'brand', etc.

I trust that everybody can see that this is ridiculous.

The reason for structure is to help guide the development of rational standards

that have some integrity of purpose and architecture. By trying to force-fit the

Ethernet MAC over the PON physical medium, you and your bretheren are going

to have to compromise away a large amount of functionallity. You are also going

to disregard many network operator requirements. This is going to produce a

You have made statements many times to the effect of, "It can be made to work".

I don't doubt it. You can patch enough glue on top of the Ethernet MAC so as to

make it operate. But that is not how systems are supposed to be designed.

Instead of straining after 802.3 compliance, we should bite the bullet and branch

off into a MAC group. That is the only way that P2MP will be able to do its work

unconstrained, in an open environment, free from the 'scope police,' who work to

shut down discussion on important issues in their unbridled quest for a quick

and very dirty standard.

You know, the flip side of this is that the PTP systems (copper and fiber) have none

of these problems. Those groups are well within the scope of 802.3, and I wholeheartedly

support them remaining in 802.3ah. Those efforts will be greatly slowed down by all

the issues raised by P2MP. All the individuals who wish to progress PTP systems

forward should think carefully about the inclusion of P2MP within 802.3ah. Now that it

has become clear that P2MP is a whole different level of complexity in comparison,

we should seriously consider again the bifurication of P2MP into a new MAC group.

Walt,

> So perhaps a time saver is to start a new 802.#(18?) where we
> don't need to spend time justifying "minimum needed" augmentation.
> .... What do you and others think?

I am opposed to this Walt. It was voted down in the past. Please let me explain why.

(1) 802.3ah EFM PAR, 5 Criteria, Objectives
We have these in place, with one year of work, and wide support. The mission is clear.

(2) Progress
I see continuous progress in defining P2MP Ethernet that offers, per the PAR, a PHY and minimal MAC augmentation. In the last meeting, it was reinforced that 802.3 has strict guidelines; I believe they can be followed without a sacrifice in system performance, and I see encouraging progress toward this.

(2) One EFM Management Platform
Local carriers will want to deploy, from a single platform under one management system, P2P, P2MP or Copper, given the particular geography. That is to say, many service providers that will deploy "Ethernet Access Networks" want to mix P2P, EPON, Copper. If you split, you lose this benefit.

(3) One EFM Hardware Platform
The consolidation of P2P, P2MP, Copper is also of benefit to EFM system vendors, who can offer systems, most elegantly, with a single backplane and packet engine. I believe the traditional Switch/Router companies follow this belief.

(4) De-focus Danger
This is the big one. When the MAC, etc, is an open door, it is much more difficult to find consensus on a solution. Tony mentioned double the years to standard for a newly formed group; I believe that is conservative.

802.x(18) EPON scope: TDM, ATM, Analog Video, DBA, QoS, FEC, DWDM, Security, Outside Plant, Numerous PMDs, newly defined OAM, .. oh, and Ethernet.
802.3ah EPON scope: Ethernet for P2MP fiber medium

I would guess that 802.x(18) would be a black hole, with no end date - not a time saver as you suggest. We can meet the timeline for P2MP Ethernet laid out by 802.3ah. We need to focus on our 802.3ah PAR, 5 Criteria and Objectives. I am discouraged to hear continuous push outside this defined envelope. Howard has built a vision for EFM that ties a common management system around a compliment of Ethernet in the First Mile solutions for carriers. I subscribe to that vision.

_________________________

Gerry Pesavento
gerry.pesavento@xxxxxxxxxxxx
Tel 530.757.6250 (Davis, CA)
Cell 530.219.1954
_________________________