Another month closer to the deadline of January 1, 2000, the day when havoc will reign in the kingdom of the microprocessor, has brought precious few new ideas for tackling the "Y2K" computer bug, the problem of the two bytes that on January 1, 2000, will think it is the year 1900.
For many hardened pessimists, that day will bring utter disaster. For those pragmatists who take the necessary steps, though, it could be just another day at the office. Most major corporations are already frantically working on this problem, hoping that time and resources do not run out. "Among [our] most important customers are financial institutions which occupy prominent positions on the Slovak financial market, and they cannot ignore this problem," said Elena Akácsová of GraTex International, a company which offers assistance in exterminating this dangerous bug.
"The solution does not lie in complicated programming, but in analyses of all software and hardware, in attempting to discover every possible place where [the "Y2K" bug] could be a problem," Akácsová reported. "Suitable solutions can only be based on analyses which take into consideration all three parameters of time, money, and risk."
Looking into "every possible place" in a computer network can indeed be a tall order that requires major investment, the kind of money that many number crunchers in finance departments will not be anxious to approve. Therefore, the problem must be seen as a top priority by the bigwigs who hold the corporation's purse before any such analyses are given the nod.
Once this important step is taken, an information technology company such as GraTex International can be set to run the gamut of steps leading to an eventual solution.
The first of these steps is called "Audit", which makes a list of all executable application programs (i.e. Load Modules, Exe Files), source codes (if applicable), and temporary fields (i.e. memory, hard drives). Finally, it creates a list of all data (current as well as archived).
Next, a solution model is designed. An average model can be broken down into the following components: 20% application design, 20% modifications, 55% testing, and 5% documentation. Currently, three technical solutions exist from which to choose - Expansion, Windowing, and Compression. Each has its pros and cons.
Expansion, the simplest solution, increases the number of digits that hold the date from six to eight (the year from two to four). This is the most logical, but not the most practical solution. There are so many places where the date is stored, including archived material, that most companies will not have the time to make the necessary alterations before the coming of the new millenium.
Windowing, also known as the "Procrastinator's Solution", is the quickest and the most pragmatic. It shifts the problem to a future date when the "Expansion" solution can be used. Simply put, the computer is forced to think that, for example, the year 1950 is the year 1900, and therefore the year 2000 becomes the year 1950.
Compression takes the two byte memory space used for storing the year and allows for the storage of four digits in a space formatted for two. This solution requires changes to the application logic and is therefore more complicated.
However, the most time and energy will go into the actual testing of the solution implementation. With testing, it is important to note the following:
1. Tests must be repeatable so results can be compared.
2. "Y2K" changes must be separate from other changes.
3. If testing is done on the same hardware, all data must be backed up.
4. While testing, setting of the date on the LPAR (Logical Partition) must be considered.
5. After testing, the computer system must be returned to its original state.
6. After setting the date to the year 2000, it is important to check if licenses, passwords, or even whole files have not expired.
Testing can be concluded when it is clear what the risks are and only when these risks become acceptable. As mentioned in a recent TechWatch column, many a database has been lost due to improper testing, and we have not even moved on to the next millenium yet.
So the choice remains: does one "fix and pay" or "wait and pray"? One thing is for sure: on January 1st, 2000, the world will see the extent of power the computer holds over us mere mortals.
Available solution tool software for do-it-yourselfers:
- Visual Age for COBOL, Professional - Professional Version includes the Y2000 Analysis Tool Module
- HLASM Toolkit - includes "Disassembler" for use with older applications
- Visual Age 2000 - An "umbrella" solution, not a complete one
- IBM COBOL for OS/390, MVS or VM - includes "Millenium Language Extension"
- Languare Environment (OS/390 component) - includes a Run-time library
