• United States
by Chris Hoogenboom

Desktops don’t need Gig E

Nov 18, 20023 mins

Gigabit Ethernet to the desktop; FCC hearings

Notably absent from Kevin Tolly’s column, Gigabit to the desktop: A decision you can’t escape, is the reason a desktop needs Gigabit Ethernet. What is the application that requires it? What is the return-on-investment model to support it? Is there even a desktop that can come close to filling a Gigabit Ethernet pipe? High-end file servers struggle to break 500M bit/sec; even if you were to run uncompressed NTSC digital video feed to the desktop, it’s only 200M bit/sec. While a Gigabit Ethernet LAN on motherboard might cost less than 10/100M bit/sec Gigabit Ethernet, Tolly presents no cost analysis on Gigabit Ethernet switch ports and backbone upgrades to support the increased load, assuming a desktop could in fact fill a Gigabit Ethernet pipe.

Chris Hoogenboom

Oxnard, Calif.

After reading about Case Western Reserve University’s investment in Gigabit Ethernet, I can’t help but wonder if the deciding factor was the imbedded fiber investment rather than an inherent need for gigabit bandwidth. Reading between the lines leads me to believe the $200-per-gigabit network interface card (fiber imbedded) was the clincher. This let the university avoid the multimillion-dollar expense for copper replacements and averted the use of media converters, which represented another unmanageable failure point.

I am very cynical about the need for Gigabit to the desktop within our lifetimes. My continued hypothesis is that human beings cannot in any way process information presented at gigabit speeds, and that any conceivable applications today require only multimegabit support. Ergo, why make the major investment in cable plant upgrades beyond Category 5 and multimode fiber when there is no compelling need?

William Dennett

Principal consultant

Enterprise Solution Providers

New York

Regarding Thomas Nolle’s column, Is it too late for FCC hearings?, here’s the real answer to the telecom problem. First, the government should buy the physical plants currently owned by the regional Bell operating companies, cable companies and anybody else that has wire in the ground. Government should then allow usage of the wire, cable, etc., on a competitive basis with very low barriers to entry and exit.

This approach also would apply to the electrical industry. The reason it works is because economies of scale say that for some services it makes sense to have infrastructure provided only once, by one entity. This works for roads; water, electricity and natural gas delivery; and sewage service. These services are provided by a government-managed or -owned monopoly because the infrastructure is very expensive to duplicate. In this environment, the company that gets in first will always have a huge competitive advantage, so we have government regulation. In addition, having multiple gas, water or electrical networks makes absolutely no sense.

I wonder what it would cost to buy all the Baby Bells, plus the cable companies?

Joe Taylor

Overland Park, Kan.

Nolle replies: I’m afraid I have to disagree. The taxpayer would pay through the nose, and there’d still be no assurance that any beneficial services or rates would arise from the deal.