Re: PMDs and Customer requirements
Walt,
Thanks for your explanation. Based on the varied PMD wishes of the reflector
traffic and the straw poll taken in Ottawa showing a majority of support for
five PMDs, it seems to me that choosing five IS the way to reach consensus and
move forward. I believe there are enough PMD guys willing to sign up to do the
work.
Regards,
Todd
Walter Thirion wrote:
> Todd,
>
> If we actually choose the 5-PMD set, the impact is the extra work in writing
> more clauses, tables, testing, etc.
>
> However, my comment was more to the issue that if we continue to try to push
> 5-PMDs without making the necessary compromises to reduce the set, then I
> think there is a large enough block of people that don't want 5-PMDs to
> block the motion. Therefore, we wouldn't be able to adopt any PMDs as the
> basis for the 1st draft and that WOULD delay the standard.
>
> Walt
>
> > -----Original Message-----
> > From: Todd Hudson [mailto:thudson@xxxxxxxxxxxx]
> > Sent: Thursday, June 01, 2000 10:28 AM
> > To: Walter Thirion
> > Cc: '802.3ae'
> > Subject: Re: PMDs and Customer requirements
> >
> >
> > Walt,
> >
> > Could you please expound on how choosing a five PMD set will
> > delay the standard?
> >
> > Regards,
> >
> > Todd Hudson
> >
> >
> > Walter Thirion wrote:
> >
> > > Steve,
> > >
> > > I understand the concern the distance objectives may not
> > properly address
> > > all of the market segments. However, it is all I have
> > (along with the other
> > > criteria of cost, tech feasibility, etc.) to try to sort
> > out the verious
> > > proposals. I remember the discussions around the objectives
> > and the various
> > > positions taken and, in many ways, we are simply replaying
> > the discussion
> > > today. But unless we revote on the objectives, I am
> > obligated to meet them.
> > >
> > > As to the straw poll, it would have been informational to
> > see it done based
> > > on representative vendor type, i.e. a straw poll among
> > equipment vendors and
> > > a separate one among PMD vendors. I have no doubt the PMD
> > vendors strongly
> > > endorse at least 5 PMDs because I know each camp is well
> > represented. I also
> > > know the equip vendors have coalesced around 3 PMDs for the
> > reasons that
> > > Steve Haddock presented.
> > >
> > > I don't think any of the positions are strong enough to
> > pass a motion with
> > > the required 75% but I think each of them is large enough
> > to block a motion
> > > that doesn't include their favorite solution.
> > >
> > > So as I said in Ottawa, we have two choices:
> > >
> > > 1. Try to compromise on a smaller set that we can pass.
> > > 2. Not pass anything and therefore delay the standard.
> > >
> > > As always, I am open to suggestions if there is another
> > path that I'm
> > > missing.
> > >
> > > My understanding is the "100m on installed fiber" objective
> > requires the
> > > 850nm or 1310nm WWDM solution. The groups discussion to
> > satisfy the "at
> > > least 40km on SMF" by having a solution that can go 50-80km
> > dictates 1550nm
> > > Serial. Since the 1310nm WWDM also satisfies the middle
> > objectives, many are
> > > happy with this simple set.
> > >
> > > Walt
> > >
> > > > -----Original Message-----
> > > > From: Swanson, Steven E [mailto:SwansonSE@xxxxxxxxxxx]
> > > > Sent: Wednesday, May 31, 2000 9:40 AM
> > > > To: '802.3ae'; 'Walter Thirion'
> > > > Subject: RE: PMDs and Customer requirements
> > > >
> > > >
> > > > Walt,
> > > >
> > > > The entire discussion around link length objectives on the
> > > > reflector to this point has been fundamentally flawed. For
> > > > reasons much different than those that Bruce provided, I
> > > > agree that the discussion thus far is like comparing apples
> > > > and oranges because they no not address the same customer segment.
> > > >
> > > > IEEE 802.3 has historically developed applications that will
> > > > run over media defined by structured cabling standards like
> > > > ISO/IEC 11801 and TIA 568. For these standards, the only
> > > > lengths of significance have been:
> > > >
> > > > * 100m for the horizontal portion of the network
> > > > * 500m for the riser portion of the network (building
> > backbone)
> > > > * 2km for the campus portion of the network (campus backbone)
> > > >
> > > > The 300m link length has gained some popularity recently due
> > > > to the emergence of centralized architectures (vs.
> > > > distributed architectures) as well as the fact that a
> > > > significant percentage of the known building architectures
> > > > could be supported at 300m.
> > > >
> > > > The development of the 10GbE Standard signalled a departure
> > > > from just supporting Ethernet over Structured Cabling - we
> > > > essentially have added three NEW customer segments that ARE
> > > > NOT addressed by Structured Cabling Standards, namely:
> > > >
> > > > * Very, very short reach for connections between switches
> > > > in the same room.
> > > > * Medium long reach ( 10 to 40 km).
> > > > * Very long reach (over 40 km).
> > > >
> > > > So, I could construct a very convincing argument that the
> > > > 100m objective over installed multimode fiber is flawed
> > > > because 10GbE IS NOT addressing the horizontal portion of the
> > > > structured cabling network. To believe this is the case
> > > > supports 10GbE to the desk in the near term which even the
> > > > die hard optimist on the fiber side would be hard pressed to
> > > > justify. And, even if it was the case, there is very little
> > > > installed base of fiber in the horizontal portion of the
> > > > network today.
> > > >
> > > > Now, what about the 300m space? I agree that an installed
> > > > base of fiber exists in this space and we have an obligation
> > > > to support that installed base. But I also strongly believe
> > > > that we should give our customers an option for addressing
> > > > that space. History shows that customers will deploy new
> > > > media if it makes sense. If not, we would still see a large
> > > > installed base of CAT3 copper in the horizontal. Yet cabling
> > > > is generally replaced if a more cost effective solution
> > > > exists for a given space. We seem to argue both ways on this one:
> > > >
> > > > * I am only going to support the installed base
> > > > * or, if I have to install new media, I might as well
> > > > recommend single-mode fiber
> > > >
> > > > Neither of these scenarios really reflects current market
> > > > dynamics. Simply put, it boils down to economics. And when we
> > > > look at the short reach application space being addressed by
> > > > 10GbE, we know that 850 nm solutions will likely provide the
> > > > economic rationale in that space. In addition, both the very
> > > > short reach and the building backbone spaces are easily
> > > > upgraded compared to the more difficult upgrade that has
> > > > taken place in the horizontal portion of structured cabling.
> > > >
> > > > We can continue to debate the merits of each of these PMD
> > > > options and hold the standard hostage or we can try to build
> > > > consensus around a solution set that has majority support
> > in 802.3ae.
> > > >
> > > > My recommendation is that we build consensus around the 5-PMD
> > > > set that by our own straw poll showed a 2 to 1 margin of
> > > > support in 802.3ae at the Ottawa meeting.
> > > >
> > > > Best regards,
> > > >
> > > > Steve
> > > >
> > > >
> > > > > ----------
> > > > > From: Walter Thirion[SMTP:wthirion@xxxxxxxxxxxx]
> > > > > Sent: Tuesday, May 30, 2000 8:55 PM
> > > > > To: '802.3ae'
> > > > > Subject: RE: PMDs and Customer requirements
> > > > >
> > > > >
> > > > > Chris,
> > > > >
> > > > > The 100m objective is a pretty standard based on multiple
> > > > surveys over the
> > > > > years for networking installations. At 100m, you can
> > > > satisfy something like
> > > > > 95+ per cent of the horizontal runs, wiring closets, etc.
> > > > Most structured
> > > > > wiring standards have, therefore, specified this distance
> > > > as maximum, so it>
> > > > > has become self fulfilling, i.e. installers specify all new
> > > > horizontal runs>
> > > > > to be less than 100m, wiring closets are placed in order to
> > > > ensure the 100m
> > > > > max, etc.
> > > > >
> > > > > The "installed base" part is relevant because of the
> > > > reasons Bruce has
> > > > > elaborated.
> > > > >
> > > > > I believe the 300m is for campus environments, building to
> > > > building, etc.
> > > > >
> > > > > Walt
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: Chris Simoneaux [mailto:csimoneaux@xxxxxxxxxxxxx]
> > > > > > Sent: Tuesday, May 30, 2000 7:47 PM
> > > > > > To: 'Bruce Tolley'; Walter Thirion; '802.3ae'
> > > > > > Subject: RE: PMDs and Customer requirements
> > > > > >
> > > > > >
> > > > > > Bruce,
> > > > > >
> > > > > > My concern here is that we are letting the needs of the few
> > > > > > drive the needs
> > > > > > of the many...so to speak. As an equipment vendor, you must
> > > > > > support any
> > > > > > application that the customer brings to you. That's a fact
> > > > > > we all would
> > > > > > agree with. However, I disagree with basing the success of
> > > > > > 10GbE on how
> > > > > > well it supports the minority applications.
> > > > > >
> > > > > > Customers are typically adverse to "forklift upgrades". But
> > > > > > in the end,
> > > > > > it's a matter of economics. And the cost vs. profit of the
> > > > > > upgrade is the
> > > > > > determining factor. If most of the user's links are
> > > > shorter than 100m
> > > > > > (probably much shorter), then the approach to the system
> > > > upgrade will
> > > > > > usually be based on the short distance link and expand into
> > > > > > the lower volume
> > > > > > longer distance links, not the other way around.
> > > > > >
> > > > > > Embedded in this PMD selection process is the 850nm serial
> > > > > > solution. It
> > > > > > clearly offers a solution for the 300m objective. It also
> > > > > > offers added
> > > > > > value of lowest cost, simplistic interconnections for all MM
> > > > > > fibers up to
> > > > > > 28m and most 50um MM fibers up to 86 meters. If technology
> > > > > > proliferation is
> > > > > > important, then 850nm serial should be a choice as it has
> > > > > > several vendors
> > > > > > that are doing actual link testing. And one that is actively
> > > > > > demo'ing in
> > > > > > the public forum. Ask that of any WDM solution. With all of
> > > > > > it's benefits,
> > > > > > why would the group not include the 850nm solution? Maybe
> > > > > > someone can help
> > > > > > me understand that.
> > > > > >
> > > > > > Also, I must respectfully ask for someone out there to inform
> > > > > > me why the
> > > > > > "100m over installed base", objective was chosen. Why not
> > > > > > 50m? or 150m?
> > > > > > Were there any scientific data or surveys used? Don't
> > > > > > misinterpret, this is
> > > > > > not a "knock" on the 802.3ae process. It's merely a question
> > > > > > that's been
> > > > > > bugging me for a while now.
> > > > > >
> > > > > > Finally, I would suggest we base our PMD solution set
> > > > > > opinions on straw
> > > > > > polls where all participants have an opportunity to vote.
> > > > > > Informal straw
> > > > > > polls can be misleading. Each subgroup within the 802.3ae
> > > > > > seems to have an
> > > > > > opinion on the PMD issue. However, we should not forget
> > > > > > about the user
> > > > > > community. The only user that stood up and voiced an
> > > > opinion (that I
> > > > > > remember) during the Ottawa meeting was MCI/Worldcom, and I
> > > > > > believe they
> > > > > > stood in support of 850nm serial. Correct me if I'm wrong.
> > > > > >
> > > > > > Chris
> > > > > >
> > > > > > -----Original Message-----
> > > > > > From: Bruce Tolley [mailto:btolley@xxxxxxxxx]
> > > > > > Sent: Tuesday, May 30, 2000 9:49 AM
> > > > > > To: Walter Thirion; '802.3ae'
> > > > > > Subject: PMDs and Customer requirements
> > > > > >
> > > > > >
> > > > > >
> > > > > > Walter:
> > > > > >
> > > > > > I have to reject the 2nd most popular set because it abandons
> > > > > > the installed
> > > > > > base of customers using their installed 62.5 nm 160 and 200
> > > > > > MHz*km fiber.
> > > > > >
> > > > > > o Buildings will not move because of 10 GbE
> > > > > >
> > > > > > o The cabling infrastructure will not change overnight
> > > > > > because of 10 GbE
> > > > > >
> > > > > > o Customers have a requirement to run 10 GbE on the same
> > > > > > fiber links they
> > > > > > run 1 GbE today
> > > > > >
> > > > > > o Customers have a requirement to use 10 GbE in
> > building backbone
> > > > > > applications on their installed fiber to distances of 200 to
> > > > > > 300 meters.
> > > > > >
> > > > > > There is only one PMD proposal that runs on the installed
> > > > > > fiber at building
> > > > > > backbone distances: the 4 lambda 1300 nm WWDM proposal.
> > > > If we do not >
> > > > > > support the installed base, it will stall the market
> > > > acceptance of 10
> > > > > > GbE. Customers resist forklift upgrades.
> > > > > >
> > > > > > At the meeting in York I felt that the study group basically
> > > > > > abandoned the
> > > > > > installed base of customer by deleting the word "installed"
> > > > > > out of the 300
> > > > > > meter goal. We have to build a standard that supports the
> > > > > > installed base
> > > > > > of building backbones. There is no good reason to exclude the
> > > > > > one proposal
> > > > > > that supports this market segment.
> > > > > >
> > > > > > I have not hung up on the number of 3 PMDs but we have to
> > > > > > have the one that
> > > > > > supports the installed base of fiber out to 300 meters.
> > > > > >
> > > > > >
> > > > > > Bruce Tolley
> > > > > > Enterprise Line of Business
> > > > > > Cisco Systems
> > > > > >
> > > > > >
> > > > > >
> > > > > > At 12:58 PM 5/28/00 -0500, Walter Thirion wrote:
> > > > > >
> > > > > > >First of all, thanks to everybody that presented PMD
> > > > > > proposals at the last
> > > > > > >meeting. I've sent my presentation to David Law, so it
> > > > > > should be available
> > > > > > >on the web site in the next couple of days.
> > > > > > >
> > > > > > >In listening to the discussion after my presentation and
> > > > > > then going around
> > > > > > >and talking to people, it feels to me like we're starting to
> > > > > > converge. Not
> > > > > > >there, yet, but making progress.
> > > > > > >
> > > > > > >The equipment manufacturers made it pretty clear they would
> > > > > > like to see no
> > > > > > >more than 3 PMDs in the standard. The PMD vendors have some
> > > > > > concern that
> > > > > > >using only 3 PMDs may sub-optimize certain objectives,
> > > > > > however, they could
> > > > > > >support the 3 PMD position if it is made clear which 3 PMDs
> > > > > > the equipment
> > > > > > >oems want.
> > > > > > >
> > > > > > >Based on an informal straw poll and anecdotal evidence, my
> > > > > > opinion is the
> > > > > > >first choice would be the set:
> > > > > > >________________
> > > > > > >850 nm WWDM
> > > > > > >1310 nm WWDM
> > > > > > >1550 nm Serial
> > > > > > >________________
> > > > > > >
> > > > > > >If that set isn't feasible, then the 2nd most
> > popular choice is:
> > > > > > >________________
> > > > > > >850 nm WWDM
> > > > > > >1310 nm Serial
> > > > > > >1550 nm Serial
> > > > > > >________________
> > > > > > >
> > > > > > >Thoughts, feedback?
> > > > > > >
> > > > > > >Walt
> > > > > > >___________________
> > > > > > >Walter Thirion
> > > > > > >Chair, IEEE 802.3ae PMD Sub-Task Force
> > > > > > >301 Congress Ave.
> > > > > > >Suite 2050
> > > > > > >Austin, Texas 78701
> > > > > > >Voice: 512-236-6951
> > > > > > >Fax: 512-236-6959
> > > > > > >wthirion@xxxxxxxxxxxx
> > > > > > >___________________
> > > > > > >
> > > > > >
> > > > >
> > > >
> >