|
Simply disciplineDeploying client/server technology is too difficult to attempt without a proper infrastructure |
Mainframe computing isn't reliable because of the hardware or software. Mainframers consult with clipboards to follow strict procedures, schedules, and disciplines. This column outlines the disciplines you should follow to make client/server computing production-quality. (900 words)
Mail this article to a friend |
If you talk to a seasoned mainframe administrator about the importance of discipline in running an enterprise computer operation, he or she will mention checklists, procedures, reviews, schedules, and accountability. (We sometimes joke that the best mainframers were born and raised in the U.S. Midwest.) On the other hand, if our experience is typical, the average NetWare or Unix administrator will blink unknowingly when you mention discipline, or maybe refer you to their favorite leather-related shop in San Francisco.
Before we take this column any further, we need to define what we mean by "discipline" in a computer-related context. Perhaps the best way is to list the disciplines we learned as mainframers and still follow today:
|
|
|
|
Mainframers should be familiar with the above. For everyone else, this list is quite a bit to swallow. Keep in mind why mainframe computing continues to thrive today, as most estimates say mainframes still support 80 percent of the Fortune 500 companies. Those legacy systems are still there and aren't going away any time soon (if ever). So why were mainframes so successful back then? And why did mainframes nose dive in popularity?
One of the reasons for the success of mainframes was due to their Reliability and Availability -- two legs on the "RAS" three-legged stool. Mainframes have their weaknesses, but they maintain 99.9 percent online access without exotic, high-availability hardware. This reliability has nothing to do with the box. So what was it? You got it, those disciplines! It took 30 years to perfect those checklists, procedures, reviews, schedules, and accountability. The NetWare and Unix crowd needs to learn what kept mainframe shops running all these years.
There is a big caveat. You cannot take the disciplines listed above and move them forward to the new environment as-is. You need to remove the bureaucracy first, streamline wherever possible, and strive for a "lights-out" data center. We'll talk more about this in in upcoming columns.
But why is it almost impossible to implement disciplines? It takes time, which of course no one has. It took mainframers decades to form the right infrastructure and culture, and now we're asking today's managers to build high-RAS client/server systems overnight. (Remember, client/server systems are three times more complex than mainframes.) So stop expecting that everything can be implemented yesterday!
On the other hand, customers are impatient and won't wait for IT to get its act together. Development cycles for new applications have reduced dramatically. Much effort is being put into developing and deploying systems rapidly. What used to take years IT now has the tools to develop in months. With the World Wide Web, some programmers are dreaming-up, developing, and deploying applications in weeks.
Is there a happy medium? Can IT offer rapid application development (RAD) with high reliability, availability, and serviceability (RAS)? Yes! But you need disciplines. You need to get some quick successes under your belt with your customers. We're talking about processes, standards, and policies that will bring IT and their customers closer together. And these processes need to be very cost efficient and streamlined.
In future columns, we will offer more details on how we've accomplished this successfully.
|
Resources
If you have technical problems with this magazine, contact webmaster@sunworld.com
URL: http://www.sunworld.com/swol-05-1996/swol-05-unix.html
Last modified: