Q: With the bus speed of desktop computers topping out at 132 megabits per second and Alpha servers at about 267 Mbits/sec, why would some large organizations want to put fiber to the desk? With the capability of Category 5 unshielded twisted-pair cable specified at Level 6 or 7, why would you change out your copper and desk- top cards for fiber at three times the expense of copper?
Do any manufacturers of desktop computers plan to develop substantially higher bus speeds? If not, isn`t putting fiber to the desk like putting server capability at each worker`s desk? Could putting fiber to the desk be compared to drinking from a fire hose, or is it a farsighted capital expenditure in preparation for emerging technologies?
Clayton Duncan
General Instrument
Lansdale, PA
A: Industry sources with whom I have spoken are seeing a migration away from intelligent workstations on every desk back to a couple of very high-end servers, serving what will be virtually "dumb" terminals at the desk. The rationale is that the network is much easier to secure and control and more economical to maintain. This is sort of an "everything-that`s-old-is-new-again" approach.
The cost of network hardware for optical-fiber media has begun to fall dramatically in recent months. There has been a small-form-factor optical-fiber connector war raging, and while no single connector has emerged the victor, the industry has benefited greatly from the alliances formed during the heat of battle. As a result, we are beginning to see reasonably priced fiber-optic transceivers become available.
I recall the naysayers when 10Base-T was the rage. What would we ever do with all that bandwidth? And here we are today awaiting a 1000Base-T standard, with the Institute of Electrical and Electronics Engineers (New York City) already looking into 2.5- and 10-gigabit-per-second applications. These applications could begin to inflate that fire hose.
But then there is the question of what flavor your fire hose is--62.5-micron multimode or singlemode, or will we see 50-micron multimode gain favor?