The following text is copyright 1996 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.

Why can't an ISP be more like a phone company?

There has been a growth industry in second guessing following the multi-hour outage of BBN Planet's POP located at Stanford University on October 11th. The basic theme of the Monday morning quarterbacks is that BBN had not been as paranoid as a phone company would have been in protecting its POP from the possibilities of power outages.

I expect that this is true. A phone company I dealt with a couple of years ago needed to replace the main power transformer for one of its central offices (CO). This CO has its own set of batteries capable of running the telephone switch for many hours and its own turbine generator capable of supporting the whole building with enough fuel to run it for days. But the phone company was not comfortable with this level of backup during the day-long transformer work. They brought in a 2nd similar-sized turbine generator on a trailer truck. That also was not safe enough. They brought in another truck mounted generator big enough to run the switch itself.

This type of paranoia costs real money. Money is quite easy to get when you are a regulated monopoly - actually in some cases, since allowable tariffs are based on costs plus a percentage of profit, the higher the justifiable costs, the higher the profit. It will be very interesting to see if the paranoia stays as high under the pending deregulation and competition.

While I can't resist saying that it might have been a good idea to have a portable generator lying around to be transported to a problem site it is a bit hard to understand in the abstract just what type of backup is worthwhile. BBN said that there was a triply redundant power feed to the site and there was a long history of very reliable power. Note that this same site houses Stanford's PBX and I expect that Stanford's own phone people tend to run a bit paranoid for reliability.

It is security 101 that a risk analysis should be done before deciding what type of security you need in some situation. It generally is not worth putting in a security system worth substantially more than the thing you are protecting. In this case one would have to take into account the failure probability and the cost of a failure. At least BBN now has a better understanding of what the cost of failure is, both to BBN and to its customers. It may help refine the risk analysis going forward.

Clearly, going forward, the quality Internet service providers must do an increasingly better job of providing a reliable infrastructure to support the requirements of commerce. But, even with higher requirements, many customers will opt for low cost over reliability and quality. This can be easily seen in the long distance telephone business. I would expect that many ISPs will slant their risk analysis a bit more to the paranoid side as a result of this incident.

An interesting thing to note though is that the new services that the phone companies are now offering do not lend themselves to ultra-reliability. With good-old analog telephone service you can pick up you phone during a wide-spread power failure and hear a dial tone. Try that with your new ISDN line -- with ISDN you have to supply local power to the phone set. If your power is out it does not help that the phone CO is glowing in the general dark.

disclaimer: Harvard has lots of ISDN phones (for voice, not data) but the emergency phones in the elevators are analog - so I can call in my personal opinions, such as this column, while sitting in the literal dark.