I recently found an excellent article in CIO Magazine regarding the risk of maintaining legacy systems: Bound to Fail. Many companies today are still running back office operations on systems that are 20+ years old. It is hard to replace a trusted application running on mostly depreciated hardware, with a new application that may require extensive capital outlays, has an appreciable learning curve, and is running on an unfamiliar platform. Change, especially change from "green screen" environments to GUI environments, is difficult. Based on my experience, changing systems is often much harder on the end users than it is on the IT departments.
Keeping a legacy system just because it runs well is not necessarily a good idea. As Jim Collins says in Good to Great: "Good is the enemy of Great". Not long ago, I experienced a near disaster with a "good" legacy system. One morning while heading to my office I made a shortcut through our Network Operations Center to check on things. As I passed a particular set of mission critical SCO boxes I noticed a programmer staring dumbfounded at a screen. I asked what the problem was. He told me the server died with a disk crash and wouldn't come back on-line. He'd called in support and was told that the system was no longer being serviced by the vendor, and the only person who had a clue about how to fix it had retired a few months earlier (no joke). I asked the programmer about the "hot swappable" backup server. He said that it had never been tested (still not joking). My reply: "It's about to be."
I'm the kind of person who enjoys rolling up his sleeves and working on problems. I booted the backup server and began reviewing the network activity. Seeing no network traffic on the server, I looked at the connections coming into the back of the box. It turned out that this server not "hot swappable"...it had never been hooked up. I rerouted the network cables from the dead server to the backup server, and the programmer and I began configuring it. After several tense minutes we were able to get the backup server up and running. The total outage was less than 2 hours.
Problem solved? No way. We were still running a system that was "bound to fail." I'm happy to report that by the end of the year the old system will be replaced. That incident, while not as dramatic as the one mentioned in the CIO article, taught me something about legacy systems. Just because a system is in place and running well for years, doesn't mean that it is without risks. As they age, computer systems become ticking time bombs. You never know when a drive will crash, a CPU will fail, or a power supply will die. My advice, make systems replacement part of your overall IT strategy.
