RE: 3PMD Proposal is not enough
Steve,
See below what I recall hearing from my customers(systems integraters) the
last time we got together as a group. If I misrepresent something, someone
please correct me.
> Jonathan,
>
> This is a great set of questions but I would like to broaden the scope to
> also include the other PMDs under consideration by 802.3ae. I understand
> (I think) the motivation for 1550 nm since support of extended link
> lengths is desired but I am wondering if others could help me understand
> the following, specifically as it relates to supporting a multimode cable
> plant:
>
> 1. What are the motivations for customer choice of 1310 WWDM? Support of
> the installed base of 300m on 62.5um MMF. Granted this is not an explicit
> objective, however it seems important to my customers.
> 2. What are the motivations for customer choice of 1310 Serial? Supports
> the existing WAN transponder implementations. Transponders here are
> defined as wavelength translaters(i.e 1310 to ITU WL in the 1550 region).
> These solutions are shipping today from several vendors and there is
> general consensus that at some point 1310 serial will be lowest cost for
> SMF.
> 3. Are they doing this to get greater distance? Customers are willing to
> pay a small premium for greater distance coverage and less PMD types to
> inventory.
> 4. Are they doing this because they believe that operating a 1310 WWDM or
> 1310 Serial system over the installed base of MMF is less expensive than
> another solution over new MMF? See answer 3. Also the issue of cost, in
> particular deltas between competing solutions, is extremely subjective for
> 10G. Depending on which side of the fence your on, the deltas between 850
> and 1310 could range from very small %'s to very large %'s. There are
> companies who ship and design modules with both technologies and they have
> made very clear choices for 10GE.
> 5. Other rationale? The VSR solutions are best served by being developed
> by forums or standards focused on serving the VSR application. These
> solutions include 850 serial and WDM, 1310 Fabry Perot, and 4 and 12
> channel VCSEL arrays. The groups evaluating these are Infiniband, OIF, and
> ITU. My main concern with doing this work in the IEEE is that we have not
> spent nearly enough time defining the true requirements of equipment room
> applications. What are the losses besides attenuation. How many patch
> panels, splitters, and other passive elements are there in the link just
> to mention a few.
>
Ed
> Thanks,
>
> Steve
> > ----------
> > From: Jonathan
> Thatcher[SMTP:Jonathan.Thatcher@xxxxxxxxxxxxxxxxxxxx]
> > Sent: Friday, June 23, 2000 7:21 PM
> > To: 'stds-802-3-hssg@xxxxxxxx'
> > Subject: RE: 3PMD Proposal is not enough
> >
> >
> > John,
> >
> > This is great information. Thank you.
> >
> > What assumptions are used? What are the motivations for customer choice
> of
> > the new MMF?
> >
> > 1. Are they doing this to get greater distance at gigabit speeds?
> > 2. Are they doing this to improve error rates at gigabit speeds?
> > 3. Are they doing this because the new MMF is less expensive than
> existing
> > MMF?
> > 4. Are they doing this because they expect P802.3ae to adopt an 850 nm
> > solution?
> > 5. Other rationale?
> >
> > jonathan
> >
> > >-----Original Message-----
> > >From: George, John Emanuel (John) [mailto:johngeorge@xxxxxxxxxx]
> > >Sent: Thursday, June 22, 2000 2:21 PM
> > >To: 'stds-802-3-hssg@xxxxxxxx'
> > >Subject: RE: 3PMD Proposal is not enough
> > >
> > >
> > >
> > >The previous message had tables that may not have been readable. This
> > >version is re-formatted.
> > >
> > >Regards,
> > >
> > >John George
> > >Lucent Technologies
> > >770-798-2432 (Voice)
> > >770-798-3653 (Fax)
> > >
> > >> -----Original Message-----
> > >> From: George, John Emanuel (John)
> > >> Sent: Tuesday, June 20, 2000 5:58 PM
> > >> To: 'stds-802-3-hssg@xxxxxxxx'
> > >> Subject: RE: 3PMD Proposal is not enough
> > >>
> > >> Howard,
> > >>
> > >> Good question.
> > >>
> > >> The short answer is that there will be more than enough
> > >installed next
> > >> generation fiber to support the approximately 2 million Ten Gigabit
> > >> Ethernet Ports that Bruce Tolley projected will ship through 2004.
> > >>
> > >(http://grouper.ieee.org/groups/802/3/ae/public/may00/tolley_1_
> > >0500.pdf).
> > >>
> > >
> > >Global MM Shipments (KMI view)
> > >2001 4,316
> > >2002 4,998
> > >2003 5,795
> > >2004 6,522
> > >
> > >% Next Gen MM of total KMI projection
> > >2001 10%
> > >2002 25%
> > >2003 35%
> > >2004 50%
> > >
> > >Next Gen MM FMM (Fiber Mega Meters) annual installation
> > >
> > >2001 432
> > >2002 1,250
> > >2003 2,028
> > >2004 3,261
> > >TOTAL 6,971
> > >
> > >Cumulative 10G ports supported by next generation MM, 0 - 300m
> > >
> > >(See assumptions below)
> > >
> > >2001 650,549
> > >2002 2,488,049
> > >2003 5,470,769
> > >2004 10,266,357
> > >
> > >Bruce Tolley (Cisco) Projection for Total 10 G Ethernet Ports
> > >(Bruce, I estimated these from your logarithmic chart, and
> > >apologize for any
> > >inaccuracies in my reading the datapoints)
> > >
> > >2001 10,000
> > >2002 200,000
> > >2003 800,000
> > >2004 2,000,000
> > >
> > >>
> > >> We estimate that between now and 2004 the installed quantity of next
> > >> generation multimode fiber will reach 7,000 FMM (fiber mega
> > >meters). The
> > >> vast majority of the next generation fiber will be installed
> > >in buildings
> > >> in links up to 300 meters. Assuming an average length of 170
> > >meters for
> > >> <300 meter links (based on the 7/96 IEEE survey), and
> > >assuming only 25% of
> > >> the fiber is lit, next generation multimode will be able to
> > >support over
> > >> 10 million 10 Gigabit Ethernet Ports by 2004. Even assuming
> > >that we ship
> > >> only half of the projected next generation multimode, a >
> > >conservative view,
> > >> we will be able to support 5 million ports.
> > >>
> > >> Bottom line: There will be more than enough installed next generation
> > >> multimode fiber to support 10 Gigabit Ethernet at 850 nm from 0 - 300
> > >> meters.
> > >>
> > >> On the installed base question in general, the global
> > >installed base of
> > >> all multimode by 2004 will be about 35,000 FMM. As of the 1996 IEEE
> > >> survey, only 20% of the installed FMM was in links up to 300
> > >meters since
> > >> most of the fiber was in the campus and building. By 2004,
> > >we can assume
> > >> the installed base up to 300 meters will grow to 40% of the
> > >total as fiber
> > >> grabs additional share of building backbones and FTTD grows.
> > >>
> > >> Thus, the installed base in links up to 300 meters will be
> > >14,000 FMM (40%
> > >> of 35,000) by 2004, and next generation fiber will comprise
> > >7,000 FMM, or
> > >> 50% of the installed base up to 300 meters.
> > >>
> > >>
> > >> Regards,
> > >>
> > >> John George
> > >> Lucent Technologies
> > >> 770-798-2432 (Voice)
> > >> 770-798-3653 (Fax)
> > >>
> > >> -----Original Message-----
> > >> From: Howard Frazier [SMTP:hfrazier@xxxxxxxxx]
> > >> Sent: Wednesday, June 14, 2000 9:09 PM
> > >> To: stds-802-3-hssg@xxxxxxxx
> > >> Subject: RE: 3PMD Proposal
> > >>
> > >>
> > >>
> > >> John,
> > >>
> > >> Can you provide some numbers to substantiate
> > >this assertion:
> > >>
> > >>
> > >> >Please note that pulling new fiber is a small
> > >fraction of
> > >> the total system
> > >> >cost. Customers have already been installing the new
> > >> multimode that will be
> > >> >used today for 1000BASE-SX, and can be upgraded to
> > >> 10000BASE-850nm in the
> > >> >future. The bulk of 10 GBE port sales will occur after
> > >> 2002. By that time,
> > >> >there will be a significant and growing
> > >installed base of
> > >> the new multimode
> > >> >fiber.
> > >>
> > >> What constitutes "a significant and growing
> > >installed base
> > >> of the new
> > >> multi-mode fiber?"
> > >>
> > >> Can you show us a timeline, going out for the next five
> > >> years, with
> > >> your projections of the percentage penetration
> > >for the new
> > >> fiber in
> > >> both the horizontal and backbone applications?
> > >>
> > >> Howard Frazier
> > >> Cisco Systems, Inc.
> > >
> >
>