re[2]: 10Gig Copper Strawman
To Copper folks,
The concept of a copper based extension to Hari is dependant on the question of if Hari can be extended. This is actually not clear.
If Hari is 'terminated' at each node, with both jitter and skew budgets being reset, then it is a violation of the Hari spec to run 20" on a PCB to a header, and onto copper. To comply, the connector will need a CMOS IC that receives the Hari, resets the jitter and skew budget, and recreates Hari.
This may seem excessive, but I've heard some interpret Hari this way.
My preference is to instead define a Hari repeater. This would be like a CDR without a clock output. In otherwords, it would accept, on a per channel basis, a Hari input and perform a jitter transfer function with a low loop bandwidth (LBW TBD). This would reject jitter above the LBW, but do nothing to reduce the skew. This Hari repeater would need to be inserted in Dan's copper_strawman.pdf on both the transmitting and the receiving sides. The equalization questions will remain.
The Hari repeater concept is critical to the x4 copper concept, as it is to WWDM or ribbon fiber.
Your comments are welcome.
Regards, Bill
>> Hi Les,
>> I appreciate your points. They align with Howard Frazier's comments at
>> the IEEE meetings. Apparently Cisco is not seeing demand for short copper
>> interconnects.
>> As for CX, I believe that there are a number of factors that limited
>> it's implementation which may or may not impact 10Gig. Initial problems
>> with EMI, and two different connectors which led to confusion.
>> There are a lot of proprietary copper interconnects today and we can
>> avoid confusion for the customer if we come up with a single, well
>> architected solution that allows multi-vendor interconnection. Unlike
>> Gigabit, we won't have a 10000BASE-T alternative to look for as a cost
>> alternative to fiber. This will be it.
>> Regards,
>> Dan Dove
>> ___________ _________________________________________________________
>> _________ _/ ___________ Daniel Dove Principal Engineer __
>> _______ _/ ________ dan_dove@xxxxxx LAN PHY Technology __
>> _____ _/ ______ Hewlett-Packard Company __
>> ____ _/_/_/ _/_/_/ _____ Workgroup Networks Division __
>> ____ _/ _/ _/ _/ _____ 8000 Foothills Blvd. MS 5555 __
>> _____ _/ _/ _/_/_/ ______ Roseville, CA 95747-5555 __
>> ______ _/ ________ Phone: 916 785 4187 __
>> _______ _/ _________ Fax : 916 785 1815 __
>> __________ _/ __________________________________________________________
>> > -----Original Message-----
>> > From: Les Poltrack [mailto:lap@xxxxxxxxx]
>> > Sent: Friday, November 19, 1999 1:53 PM
>> > To: DOVE,DANIEL J (HP-Roseville,ex1); '10GigCopper'
>> > Subject: Re: 10Gig Copper Strawman
>> >
>> >
>> >
>> > Dan,
>> >
>> > This is an interesting idea, but as a "marketing guy", my
>> > sincere opinion is that the market for 10Gigabit over
>> > 16-conductor copper cable is very close to zero.
>> >
>> > A closer analogy would be 1000BASE-CX, rather than 1000BASE-T.
>> > 1000BASE-CX also required a less than common cabling scheme, and
>> > has not achieved a high-volume position in the market.
>> >
>> > I have had literally zero users ask for 1000BASE-CX support,
>> > and I have talked to hundreds of users about Gigabit Ethernet
>> > technology, products and applications. I anticipate the
>> > same behavior for 10Gig over a 16-conductor Cu cable.
>> >
>> > What customers value even more for this class of application
>> > is compatibility, which 1000BASE-T delivers, and neither
>> > 1000BASE-CX nor this proposal provide.
>> >
>> > Best regards,
>> >
>> > Les Poltrack
>> > Product Line Manager, Gigabit Ethernet
>> > Cisco Systems, Inc.
>> >
>> > At 03:34 PM 11/18/99 -0700, DOVE,DANIEL J (HP-Roseville,ex1) wrote:
>> > >Hi All,
>> > >
>> > >I thought I would start the ball rolling for a technical concept by
>> > >posting this strawman. If someone finds a serious problem
>> > with it, please
>> > >respond. Cable vendors.. post IL, RL and NEXT curves.
>> > Connector vendors..
>> > >do the same. Silicon vendors.. does it look feasible?
>> > >
>> > >In the mean time, I am soliciting feedback from system
>> > vendors and server
>> > >companies to provide a breakdown of copper versus fiber
>> > links for their
>> > >high-speed switch connections. If you have data to share, please do.
>> > >
>> > >As I stated in the meetings, customers prefer copper when
>> > they can save
>> > >money. We have seen a HUGE response to our recent 1000T
>> > introduction by
>> > >customers who grudgingly used SX to attach our switches
>> > together in the
>> > >wiring closet. I expect that as servers get 1000T NICs, we
>> > will see a
>> > >similar movement to copper in those applications. To
>> > quantify a bit, I am
>> > >getting an HP marketing guy to put together some slides to
>> > provide more
>> > >solid numbers to back those statements up.
>> > >
>> > >Let's get this ball going!
>> > >
>> > >Regards,
>> > >
>> > >Dan Dove
>> > >
>> > >___________
>> > _________________________________________________________
>> > >_________ _/ ___________ Daniel Dove
>> > Principal Engineer __
>> > >_______ _/ ________ dan_dove@xxxxxx LAN PHY
>> > Technology __
>> > >_____ _/ ______ Hewlett-Packard Company
>> > __
>> > >____ _/_/_/ _/_/_/ _____ Workgroup Networks Division
>> > __
>> > >____ _/ _/ _/ _/ _____ 8000 Foothills Blvd. MS 5555
>> > __
>> > >_____ _/ _/ _/_/_/ ______ Roseville, CA 95747-5555
>> > __
>> > >______ _/ ________ Phone: 916 785 4187
>> > __
>> > >_______ _/ _________ Fax : 916 785 1815
>> > __
>> > >__________ _/
>> > __________________________________________________________
>> > >
>> > >
>> >