In 1991, the library bought its own central computer equipment -- the first IBM RS 6000 in Colorado. This machine later become a business standard.
Before then, we were spending some $80,000 a year for computer support. The total cost of the new computer (including software and data transfer) was just under $100,000. We've spent about $20,000 a year for support ever since.
That was seven years ago. Thus the total cost for the system has been $240,000. If we'd stayed with our previous system, we would have spent well over twice that much. So our decision saved a quarter of a million dollars of taxpayer money.
But alas. While seven years of service by a single computer is pretty good, as these things go, our machine is starting to show its age. Most particularly, the older computer won't be able to adjust to the dread Year 2000 problem.
For those of you who aren't up to speed about this, here's the issue. Big computer systems really didn't come into their own until about the mid-nineteen-seventies. At the time, dates were expressed as just six digits: two for the month, two for the day of the month, and two for the year.
It didn't seem to occur to anybody that in 25 years, two digits wouldn't be enough to account for a new century.
For the modern PC, that didn't present much of a problem. Most PC's last but two or three years in a household or business. And most PC's are used for word processing, not for databases involving lots of date calculations. (Incidentally, Macintoshes don't have the problem at all -- yet another sign of the foresight of its designers.)
But for the many mainframes and minicomputers that store the business records of America, the year 2000 presents all kinds of difficulties. Some data records refer to customer birth dates -- 07/13/04, for instance. Ten years from now, it might matter (to Social Security offices, for instance) whether someone were born over a century ago, or just a few years earlier. For libraries, the issue comes down to other date-sensitive information. Suppose you checked out a book due December 31, 1999. But because of various perfectly natural circumstances, you find that you can't get it back until January 2, 2000. Our system might calculate 100 years of fines. (Hey, wait. Why do I want to fix this?)
Frankly, the whole situation is ludicrous. The problem is software. How hard can it be to add two digits to a single data field?
But the solution is hardware. The software fix depends upon the presence of new processors. The new processors come with new hard drives, and a host of other protocols and peripherals. In short, the "fix" is good for business. Not library business, you understand, but computer manufacturer and retailer business.
I bring all this up not because of the inherent human (and economic) drama of the situation. I bring it up because our best solution to a problem that won't surface for two more years is to replace our computer equipment today. This means a worst-case situation of 2 weeks of downtime: one week (June 1-June 5) to upgrade our hardware, and another week (June 22-June 26) to upgrade software. Things might go better than that, but it's best to be pessimistic.
We'll set things up so that nobody has materials due during these periods. We'll still check things out to you, but the computer catalog may well be down while we're transferring files and rebuilding indexes. And we won't be able to check anything in until we come back up.
By the end of this process, there will be some good news. Those of you with Internet access will be able to search the library's holdings right from your browser, "clicking" rather than typing menu choices. You'll also be able to renew your holdings from home, even in the middle of the night.
We'll have other good news, which is fodder for future columns. Everything should be a little speedier.
Meanwhile, I extend my heartfelt apologies for your inconvenience, which we will do our best to minimize.
No comments:
Post a Comment