|
|
|
|
|
|
|
|
by Peter H. Salus
There are many folks who worry about planes falling out of the sky and civilization ending at 00:00:01 on January 1, 2000.
Of course, this is the date only for Western Christians. For Eastern Christians, the calendar is shifted by about two weeks. For Jews, it's about 240 years until the (next) millennium. For Moslems, it's also several centuries. So, the whole to-do seems both parochial and silly. Also, for computer folk, there's the realization that the calendar isn't a UNIX system: it doesn't begin with ``0'', so the second millennium must begin on January 1, 2001. (If you think there was a year 0 between 1 BC and 1 AD, this will be where you stop reading.)
However, there are problems, and they are worldwide. Why? Because over the past 50-odd years, chips and computers have become ubiquitous. Computers don't actually think very well. Unlike bookkeepers from the Italian Renaissance to the present, who used two-digit dates and had no problem computing interest for 1695 to 1705 when they wrote ``95-05'', computers don't really know how to handle ``00.''
Well, in most cases we just don't know. It appears that the overwhelming portion of the U.S. power grid has tested okay. But what if a very small supplier ``trips'' something? The grid folks assured those of us living in Manhattan in 1965 that a massive grid failure couldn't happen again. There was another less than a decade later. A few years ago, there was a cascading failure on the West Coast. So we really aren't sure what will happen.
On the other hand, there are myriad things each of us can test: kitchen and household appliances with date chips, for example. I tested my VCRs, setting the date to December 31, 1999. In the morning, one had the right time, one was blinking ``00:00'' at me.
Maybe my coffee maker won't go on, or my bread machine, or the VCR won't tape the nth episode of something. But none of these is necessary to my life.
I have confidence in the various phone companies, too. They have far too much at risk not to have checked things out. And I know that most electronic switching systems are Y2K safe.
Doors with time locks may not be. I don't know whether banks have checked the locks on their doors and their safes as rigorously as they have checked their accounting machinery. But I know the New York Stock Exchange is in good shape, as are the large brokerage houses.
If you're running UNIX, time doesn't run out for decades; well, at least into the 21st century for Linux. DOS 6 and earlier? Win95? You may be a victim. Windows 98? Well, according to Agence France Presse last February, the French Directorate for Competition and Prevention of Fraud (DGCCRF) showed that Windows 98 and Works 4.5 would be unable to recognize the year 2000. According to Marylise Lebranchu, the Minister for Small and Medium Businesses:
The DGCCRF carried out tests in mid-January on products which might not work after 2000 and we have proof that Works 4.5 and Windows 98 will not work. It is extraordinary that a company which is supposedly at the cutting edge of technology has sold products which will not work after 2000.
One of the design basics in the ARPANET, the sire of the Internet, was that it be both redundant and resilient. Today's Net, with over 50 million host machines and about 150 million users, is even more redundant. Many of us ``live'' on a frail branch, a mere twig, a single leaf. We depend upon an ISP--an enormous one like UUNET or a small one with perhaps 100 to 200 clients. Many of these smaller ISPs run older routers.
Go look at Cisco's site. All the routers currently sold are Y2K sound. Go back a couple of years: either all are sound, or there is free software available for download over the Net. If you go back five years, not only are the routers not Y2K-safe, they don't have enough memory to be upgraded.
So I worry that Internet users at some small twigs may lose their dial-up connections. I worry that some DNS caches may fail or be corrupted. Most of all, I worry about people's reactions to not being able to dial up, not being able to see the ball drop in Times Square from their desktops (as opposed to their TV sets), or not being able to exchange chat New Year's greetings.
If one looks at the Net statistics for hurricane Floyd last September, it's clear the usage spikes came after the hurricane passed: folks wanted to know ``How bad was it?'', ``How's uncle Fred?'', ``Are you okay?''
So the big Y2K events may occur before noon in the continental U.S. on December 31. When it's midnight in Guam, it's 8 AM in New York and 5 AM in Silicon Valley. In the US, we'll have a lot of warning: from Tokyo, Melbourne, Singapore, etc., as the hour marches around the globe.
I don't think we should worry. There may be glitches and minor failures, but it won't be the end of the world. Penguins will survive.
Peter H. Salus, the author of A Quarter Century of UNIX and Casting the Net, is Editorial Director of SSC.