"S*** Happens, Even to Cisco, at&t and Apple

Duke University's campus Wi-Fi network reported was being flooded by Apple iPhone MAC address requests, temporarily knocking out anywhere from a dozen to 30 wireless access points at a time. Turns out that isn't the case. It was a powering issue. Good news for Apple, as the iPhone isn't the culprit at all. Still, the outages are a reminder.

For those of you who continue to think communications infrastructure is easy, this is a reminder that "stuff happens," all the time, in unexpected ways, to the "dumb pipes" we all depend on. I just got a new Linksys Wi-Fi router to hook up to my Covad T1, for example, and though the install wizard was really nicely put together, the Linksys would not talk to the Cisco router.

It is supposed to be so easy there is no indication anywhere in any of the documentation about what Web site to go to, or what support number to call, in case installation failed, which it did, repeatedly. I finally realized I was going to require tech support so figured out where to get that from Linksys. The IM support system worked fast, and well. The connection is up. But not before reinstalling the software load.

I recall remarking to the Best Buy salesperson that I didn't have any questions, and wouldn't need any help, because I expected the hardware choice and install to be "drop dead simple." That clearly is the way Linksys designed the system, and I suspect it almost always works. Unfortunately, in this case we had to reinstall the software.

The Covad install took "longer than expected" because we were getting unexpected packet loss. To make a moderately long story short, it was a physical media failure on a short jumper in the network interface unit. Go figure. That's the last thing one would expect from new wiring.

The point is, even well designed consumer interface procedures, such that put together by Linksys, Cisco, Apple and Covad, will fail on occasion, for all sorts of apparently odd reasons. Nothing is always drop dead simple, even when well-designed processes nearly always have that intention and result.

Just because we use "dumb pipes" to some extent does not mean the networks are not occasionally "surly" and prone to failure. Far from it.