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

Re: [802.3BA] 802.3ba XR ad hoc next step concern



Geoff and Bruce, I don’t think the car story is a valid analogy.                                                          

 

As we go from 10GbE to 100GbE, there is an evolutionary transition in infrastructure – a larger portion of the link distance must and will be cost-effectively covered by MM fiber, but the 802.3ba objectives don’t reflect this.

 

For 10GbE 10G-T, 802.3an had an objective to support up to 55m (class E) and 100m (class F) on copper.  http://ieee802.org/3/an/public/mar04/agenda_1_0304.pdf

But for 100GbE 802.3ba we now have an objective for 10m on copper http://ieee802.org/3/ba/public/jul08/agenda_01_0708.pdf

 

Thus for 100GbE the 10-100m length must now be covered by fiber. This can be done in the most cost-effective way by taking the OM3 fiber and 850nm lasers used for 10GbE for 100-300m applications and streamlining them for this application. This is implicit in the existing MM objective.  The full capability of 10GbE 850nm VCSELS and OM3 fiber is not needed for most if not all of this length range at 100GbE, since they were designed for 300m in 802.3ae, and cost reduction opportunities exist.

 

However, a portion of the 100-300m range covered by OM3 fiber in 10GbE will still be needed for 100GbE applications in data centers. Fiber suppliers such as Corning are aware of the market demand in this area (as are other suppliers) and presentations have been documenting the use of longer lengths in data centers. http://www.ieee802.org/3/ba/public/AdHoc/MMF-Reach/swanson_xr_01_0608.pdf   . The OM3 MM solution is the most cost effective way to address the 100-200m range.  A target distance of 150m is ½ of the 300m 10GbE capability and can easily be accommodated. However, it is best done with a separate objective allowing the short length MM to be stripped down to absolutely minimize cost.

 

In analyzing the distribution of lengths in data centers and high performance computing environments, it is a fact many of the short links are currently copper.   It is a mistake for us to think that because the shorter lengths will now be MM fiber, that there is no need for the longer lengths currently supplied.  This mistake is an artifact of how the data is being presented.   

 

It is absolutely contrary to the interests of the end users for 802.3ba to neglect the 100-150m length segment with a simple low-cost OM3 MM solution.    

 


From: Bruce Tolley [mailto:btolley@xxxxxxxxxxxxxx]
Sent: Tuesday, October 21, 2008 2:25 PM
To: STDS-802-3-HSSG@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3BA] 802.3ba XR ad hoc next step concern

 

Geoff and Steve,

 

For customers wanting TP links further than 100 meters, I thought we pointed them to ISO 11801 and TIA568B which recommend fiber :)) for such runs.

 

Bruce

 

 

 

 


From: Geoff Thompson [mailto:gthompso@xxxxxxxxxx]
Sent: Tuesday, August 26, 2008 11:40 AM
To: STDS-802-3-HSSG@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3BA] 802.3ba XR ad hoc next step concern

At 06:35 PM 8/25/2008 , Swanson, Steven E wrote:

What should we tell our customers who have link lengths longer than 100m and want (or require) a standardized solution?


We should tell them the same thing that we tell our twisted pair customers.


That would be that their market is not large enough to:
  - Warrant a separate solution in the standard.
  - Burden the 100 meter solution with the extra costs required to meet their needs


There is nothing terrible or onorous about this. A line has to be drawn for all volume products. For passenger automobiles, the standard is clearly 5 passengers. Does a car salesman have to make excuses that the bulk of his product line only has a 5 passenger capacity? Of course not. Do some car companies think they can make money with larger vehicles? Of course.