Re: TECH: Wireless gets real

From: Eugene Leitl (Eugene.Leitl@lrz.uni-muenchen.de)
Date: Wed Jul 04 2001 - 08:03:54 MDT


On Wed, 4 Jul 2001, Mike Lorrey wrote:

> Sure, but how many years is that fiber good for? 10? 20? Figure how many

There's fiber, and there's fiber. Monomode is nice, but the new hollow
core fiber promises to be able to lick the NLO effects setting an upper
limit to fibre's capacity.

I'm still surprised that fiber is not seen as a basic infrastructure, such
as gas, electricity, canalization. The costs of the fiber itself is
relatively insignificant, if compared with the earthworks, and the people.

> years of Moore's Law-like doubling of bandwidth/utility demands maxes

Bandwidth grows runs rings even around Moore. I hope we'll be getting
hollow fibre and purely photonical switches soon.

> out the capacity of the fiber. Take that many years, divide it by 100 to
> get your discount rate. Subtract inflaction from the discount rate for
> your adjusted discount rate. Divide the cost by the years and then
> discount each year by the adjusted discount rate amortizing the previous
> years fraction.
>
> I freaked out the upper management at Dartmouth College a few years ago
> when I was helping to install their new campus wide 100 base t network.

CAT 5 or CAT 7? Even with copper, there are differences. Of course, right
now I wouldn't lay copper anywhere, unless budget limited. Fiber, at the
very least multimode fiber (good for up to 10 GBps at short range).

> I was installing the hookups in the President's office and chatting with
> his secretary, when I casually mentioned,"Yeah, you guys won't have to
> replace this system for about 10 years." which freaked her out.
> "This is supposed to be good for 20 years," she said.

The speed of the network is not terribly important, once you're past a
certain threshold. Of course, that threshold is pushed back continuously
due to the protocol bloat..

> "Nah, you gotta account for Moore's law, see, where bandwidth doubles
> every year and a half. Right now you guys have a 1 mb/sec network that's
> maxed out. Double that every year and a half, and in 10 years you are at
> 100 mb/sec which is this new systems maximum capacity." Apparently my

If they'd use CAT 7 instead of CAT 5 they could upgrade to GBit Ethernet,
when the NICs and the switches become more affordable (right now they're
an order of magnitude more expensive, and the switch costs/port are even
worse).

> employer had not either accounted for this doubling, or else had not
> told the College (and apparently the college either wasn't aware of it,
> or those who were aware of it were keeping their mouths shut, expecting
> such info to not get out until the network was bought and installed....)

For many small companies, 10 MBit coax will still do. Anything is better
than sneakernet.

-- Eugen* Leitl <a href="http://www.lrz.de/~ui22204/">leitl</a>
______________________________________________________________
ICBMTO : N48 10'07'' E011 33'53'' http://www.lrz.de/~ui22204
57F9CFD3: ED90 0433 EB74 E4A9 537F CFF5 86E7 629B 57F9 CFD3



This archive was generated by hypermail 2b30 : Fri Oct 12 2001 - 14:39:41 MDT