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

More notes from the field.

Title: RE: 850 nm solutions

I think Rich brings up a valid point.  Sometimes it seems that the engineering of these technologies is a bit out of sync with their real use.  I can remember hearing just a few years ago that GigE would only be used for major backbones and servers were not likely applications for that much bandwidth.  After all, who needed a connection faster than FDDI or OC-3 anyway?  Issues were raised about interrupts, checksums, backplane speeds, etc and the majority consensus seemed to concur. 


It sure appears to me that bandwidth is consistently like closet space, if you have it, it gets filled up.  For example, we are today using GigE Jumbo to move ~1TB/day @ 50-100MB/sec between just two machines for various backup/disaster recovery purposes and it is growing by ~50+GB/month.  This is just plain old accounting data on 1999-class Unix machines with ~12Gbit/sec of I/O capacity.  We are a not a large user of bandwidth when compared to most other similarly sized companies I am aware of.


The only reasons we have any CSMA/CD in our Ethernet world are:


  • The device is too old or low-cost to support Full-duplex.
  • We need to get a better look at the media with a protocol analyzer.
  • It is too difficult to manage the logistics of which ports are 10 versus 100, full or half duplex since Auto-Neg does not work reliably at all.


We would run 99% of our connections Full Duplex if we could.  In our case the need for new links with CSMA/CD has pretty much vanished.  I should think that since there are no real uses for CSMA/CD in GigE,  that this implies that 10GigE has basically similar requirements.  These are now just point-to-point data pipes with some form of single access flow control and classification system (VLANs, Channel Aggregation, QOS, etc).  This is the new Ethernet and I would think that upward compatibility is the right path not total upward and downward compatibility.  So what if it is not "Ethernet" anymore.  I am glad my current vehicle doesn't use carriage lamps for headlights and that it no longer resembles a horseless carriage.  It is still my means of transportation.


 I do not need to agree personally that a feature is useful to me, but if it is useful to a significant number of users, then at least let them have a standard way of implementing each of them.  I believe that Copper 10Gig and Jumbo Frames and their cousins deserve a standards based implementation plan for the market to use and let that same market choose the features they need.  This will determine the next step in Ethernet networking evolution.  The IEEE I would think, is called upon to act as a guide and overseer that can keep the market out of the gutters when it comes to technology, but not as the decision maker for what the business world wants, anymore than the business world would want to define all the specs for the encoding schemes or interconnect pathways.  It seems to be a necessary partnership.  Without businesses, there would be no need for these devices and few funds available to create them.  Without IEEE and all the engineers, we would still be using the Pony Express and quill pens.


OK, so maybe I am oversimplifying things a bit, but I believe a robust workable standard with a few extra definitions we might not use the way they were anticipated is better than a religious or minimalistic approach where we are bumping in the to the lack of standardization as an obstacle, while clinging to the past work of others.  I am constantly amazed at the technology we have today, but I would think we should strive improve much of it as we learn.  Since I cannot predict the future it seems wiser to leave it open enough to not get in the way without allowing such lax controls that anarchy reigns.



More ramblings from the Midwest,


-Corey McCormick






 -----Original Message-----
From:   Rick Walker [mailto:walker@xxxxxxxxxxxxxxxxx]
Sent:   Wednesday, April 19, 2000 6:58 PM
To:     stds-802-3-hssg@xxxxxxxx
Subject:        Re: 850 nm solutions

> Jim Tatum writes:
> But why does it matter? Why limit the users? Why not put in the table. It
> costs nothing.  Just put in what the model and data tell us to.  It is
> my opinion that a large percentage of 10GB style links are going to be
> very short, less than 10m.  If you look at the way many fiber ports
> are being used today, many are in the 10m range.  Also, since copper
> cables are going to be EXTREMELY challanged to go that distance at
> 10GB, why not let the market choose the lowest cost solution using
> 850nm VCSELs and 62.5um fiber?

FWIW, I agree that 10G across CAT-6 or other twisted pair would be very
difficult.  However 10G across coaxial cable is fairly easy.  It can be
done with 0.1" diameter coaxial cable using simple NRZ data encoding.  A
simple FIR pre-equalizer can double this distance.  Without a doubt
copper would be the cheapest solution for links under 10M.  I would
estimate a mature chipset price of about $50 per end and $15 for the

This performance was demonstrated in 1998 using a 25GHz bipolar chipset.
See: Walker, R. C., K. Hsieh, T. A. Knotts and C. Yen, "A 10Gb/s
Si-Bipolar TX/RX Chipset for Computer Data Transmission" , ISSCC Digest
of Technical Papers 41(February 1998), 302,303,450.

A Copper PHY was voted down by the committee because it was thought that
there was no market for this type of low-cost short distance link.

kind regards,
Rick Walker