All, As Brad observes, there are things that are more important for broad market potential than exactly 5 m. 1. Such a port must be compatible with the longer-reach alternative (an active optical cable like 100GBASE-SR4). Because 5 m is more onerous than needed for an intra-rack link and not enough to be sure to connect between racks. If some racks to be connected are within this range, others will not be, so any switch product that connects outside its own rack must support either just the longer-reach option (e.g. active optical cable), or both options. 2. Must allow for adequate loss in the host. See slides 7-8 of http://ieee802.org/3/ba/public/jul08/dawe_03_0708.pdf and see http://ieee802.org/3/ba/public/jul09/gustlin_04_0709.pdf . 3. Must allow reasonable cable thickness, bend radius and weight, as Brad observes. 4. Must allow backward compatibility with 10G/lane nPPI and 40GBASE-CR4. 5. Must use the same silicon as for backplane and active optical cable without adding extra burden (cost, heat, complexity). Therefore copper cable reach should not be an objective (a contract with the Working Group) to be set in advance, but the consequence of other factors. It should be the difference between what the silicon for backplane can do, and what the host itself needs for loss and crosstalk - which is how it worked out in the end for 7 m of 40GBASE-CR4. A 5 m cable objective is unhelpful - it directs a Task Force to optimise for the wrong thing. Piers From: Brad Booth [mailto:bjbooth@xxxxxxxxx]
Sent: 06 June 2011 18:59
To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_100GCU] Successful May Interim I did look at bugg_01_0511 prior to my comment. The presentation only contained IL data, not bend radius data. My statement about 22 AWG was only an example that if the IL required for a 5m twinax copper cabling reach could only be satisfied by a 22 AWG twinax cable, that while the electrical properties would satisfy the objectives the bend radius and probably diameter of said cable would be unlikely to have broad market potential. Brad, My apologies for not replying sooner, as I had limited email access this week due to travel. There was a considerable amount of discussion on this issue, which I won’t rehash. One thing that I would say that was left out is that if you look at bugg_01_0511, data based on 26 AWG and 30 AWG has been provided to the group for consideration. Therefore, I would say that the conversation at the March meeting did have an impact, as I am sure we could agree that the IL data would improve if 22AWG had been provided, but it wasn’t. Regards, John Glad to hear that the meeting was successful. Was there any consideration for listing the gauge of the wire for the 5m twinax copper cabling objective? At the meeting in Singapore, there was a request during the straw poll to list the gauge of wiring and it did slightly alter the results of the poll. Even the presentation by Mark Gustlin at last week's meeting highlighted the importance of bend radius and diameter which can be directly related to the gauge of wire. Did the study group discuss the bend radius and diameter requirements? Is it fair to assume that the gauge of wire for the twinax copper cabling objective is the same used in the Singapore straw poll (24 AWG)? If 24 AWG is assumed, was there any information presented or provided to indicate that the cable would be able to meet the bend radius, diameter and weight requirements for the intended application? If there is no information provided that indicates 24 AWG twinax copper cabling can meet the bend radius, diameter and weight requirements for inter-rack and intra-rack interconnect, then has the study group sufficiently responded to broad market potential? On Sat, May 28, 2011 at 11:39 PM, John D'Ambrosia <jdambrosia@xxxxxxxxxxxxxxx> wrote: Dear Study Group Participants, I hope everyone had safe journeys home. I have to say it was one of my more adventurous travels. However, given the success of the meeting, the travel was well worth it! The Study Group was successful in reaching consensus on the objectives, PAR, and 5 Criteria responses. They have been posted, and may be found at http://www.ieee802.org/3/100GCU/index.html. July will be the next key milestone, as we look forward to the necessary approvals to move from Study Group to Task Force. The minutes will be posted shortly. Best Regards, John D’Ambrosia Chair, IEEE 802.3 100Gb/s Backplane and Study Group
|