When Mr. Winchester passed on to his reward, Mrs. Winchester commissioned the construction of an addition to their home in San Jose. Once started, the good lady could not stand the thought of firing the construction crew so they just stayed on the job ... for about 30 years.
The result is a rambling unplanned structure that sprawls over a huge parcel of land. The house features stairways that terminate in the ceiling, hallways that wind around going no where, windowed towers and spires with no access, chimneys that terminate at the roof, doors that open into brick walls, rooms within rooms and so on. The place is bizarre an unplanned, barely functional, inefficient octopus of a house.
This is the systems model that has developed over the years in our industry. Build, connect, adapt, connect, buy, connect, adapt, connect and so forth. We must learn to work with this and ultimately perhaps to replace it with something better; but first we have to work with it as is.
What We Need
What the IT industry needs is not just a new crop of people with older skills. What it needs is a genuine culture change that promotes the concept of knowledge sharing and a culture that recognizes that those systems supporting our enterprise year in and year out have immense value. Further, developing those systems must include more than bolting on new modules built in current languages. Whole systems must evolve, not just connect this part or that layer.
So, the problem is much bigger than just COBOL and a dearth of COBOL programmers. How do we approach this? What gets the knowledge widely dispersed? What gets people thinking about the whole instead of some tiny part?
We have learned two valuable lessons in the IT area over the last few years: First, we cant allow the IT group to insulate themselves from the rest of the enterprise. They cannot be locked away in some technological equivalent of a monastery creating systems and solutions on their own, without external input, pushing these out and into the enterprise.
By the same token, total control and authority for IT strategy and implementation cannot be decentralized into individual departments or group management. Both of these approaches have failed to deliver successfully for the enterprise.
What is needed is centralized administration of the IT operation with distributed responsibility for ultimate oversight and direction. The second part should allow input from assorted elements in the user community, upper management and major functional groups within the enterprise.
Technology should be selected based on several organization-friendly elements including: adaptability, scalability, compatibility, and traceability meaning the technology can be documented in a way that future generations can understand how it works and make periodic adjustments as needed.
Finally, I think IT management has the responsibility to facilitate communication and skills training among all IT employees. If your lunchroom has a group of old-timers sitting around one table while the kids hang out at their own table, youve got a problem.
Have some contests, create teams that are made up of kids and geezers. Have knowledge-exchange seminars. Put the most successful teams to work on your most critical projects.
This is how you change the culture that got you into a corner in the first place. You will be developing a self-sustaining system that will not obsolete itself, but rather fix itself as a natural part of its own evolution.
Lou Washington is the master of MIPS at software and services provider Cincom Systems. In his spare time, hes also a senior business manager. He can be reached at firstname.lastname@example.org. Nearly one-third of Cincoms customers still use its products running in the mainframe environment.