Pay Me Now, Or Pay Me Later
Let's talk a bit about practicality. There are likely many people out there who are driving cars which have seen better days. I'm not talking about beat-up wrecks, just cars that are not the latest models and styles. Aside from the minor "dings" the bodies are in good shape, and as top heavy as their odometers may be, they still purr like kittens when you turn that key to start them. Acknowledging the fact that a car is often the second largest investment a person makes (with a house being the first), it's only logical that we want to get the most out of our investment. Why sell the car when it still runs fine? Why take on the burden of a new car payment when the "old girl" can still get around? Logical yes, but is it practical?
Age takes its toll whether we're talking about a car, a washing machine, or a computer operating system. If your fuel pump fails you can usually locate a replacement. The options open to you at that point are to either replace the pump yourself, or take the car to a mechanic with a somewhat higher degree of experience and dexterity. If you opt for the mechanic as I would, there are realities which must be faced. While I don't mean to slam mechanics, people like myself are pretty much at their mercy when it comes to the cost involved to repair outdated products. While a fuel pump may not be the best example to site here, consider the possibility of special tools and knowledge that may be required to do the job right. Supply and demand usually dictate price in these situations, and you just may find that there will come a time when it just isn't worth coughing up the bucks to
keep the vehicle any longer. What should also be considered is the possibility that the day may come when the needed parts are no longer available, or the mechanics are no longer able to affect the repair. In either event, I think it's safe to assume that the older a product becomes, the more costly it's likely to be to maintain it.
Computer software is no different. Newer versions or releases of software have proliferated the data processing industry. Certainly this is most evident in the PC arena, with new versions of DOS, Windows, etc. being announced practically monthly. Although the frequency of upgrades in the area of mainframe operating systems is not as recurrent, it does happen, and it does present the user with some problems. First of all, there are the costs involved in upgrading. These range from the obvious issues of time and manpower resource, to the not-so-obvious issues of upward application compatibility, architectural changes, and even training. Along with these factors comes the inevitable announcement from the vendor that your current software will no longer be supported at some date in the near future. So what are the choices here?
You can bite the bullet and move forward with upgrading your system to the new release, or you can choose to delay the upgrade and continue to support your system by yourself. And there are some shops that do choose the latter. After all, there's nothing actually wrong with the system today. It still runs fine, availability is still 99.8%, and any shop with an experienced Systems Programming group can deal with the dumps and problems that may crop up. So what's the problem with not upgrading? The problem with not upgrading is the ever-present risk you run of being exclusively responsible for your system. New releases of software are not just cases of additional functionality. New releases of software contain fixes to existing problems, and frequently reflect complete rewrites of processing logic to provide greater operating efficiency. Ignoring the opportunity to upgrade is not only a foolish business decision, but a decision that will ultimately cost the company more money than the upgrade itself. Consider the potential turnover of personnel who wish to remain technically current and marketable. Can a company afford to lose the very people it intends to rely on to maintain their "unsupported" system? Consider the additional time it may take to fix a problem when there is no vendor to turn to for PTF's or on-site support. Oh, IBM hasn't completely left you on your own. On a consulting basis they are always ready to lend a hand at $175.00 or more an hour. Independent contractors are also ready to respond at a price. But these things aside, is the greater matter of your competition.
Every Ford has it's General Motors. Every United Airlines has it's American Airlines. Every Prodigy has it's Compuserve. What really is the cost to a company when the competition has the technological edge? Part of the cost can be measured in potentially lengthy unscheduled downtime, but what about the inability to satisfy the needs of your customer? When you can't install TPFDF or C/370 because you're running on a 2.3 system, your application development time becomes a factor. When you can't provide 4K support because you don't have sufficient working storage below the 16meg line, your performance becomes a factor. When you can't communicate with other databases on other processors, your ability to access the information you need becomes a factor. All of these things directly effect your ability to serve your customers competitively. Add to this the likelihood of in-house modifications to fill these gaps, and the ease of eventual migration to a current release also becomes a factor. Every one of these items costs a company money, either now or later.
Running on unsupported software is not a shrewd business move. It's not just a question of paying for an upgrade now or later, but a question of how much you will pay for that upgrade. Does it really make sense to defer an upgrade when the financial benefit of acting now is obviously the smart thing to do? While none of us has money to burn, some of us do choose to spend that money wisely and practically. The rest are probably out looking for a good mechanic.