Best Dressed for Armageddon

In 2006, the world learned an inconvenient truth. The planet we call “home” was getting hotter and we were to blame. Naturally, being intelligent and rational beings, we took the action necessary to prevent the oncoming catastrophe. No? Surely then, being intelligent and rational beings, we prepared for what was to come by stopping building houses on barrier islands. No?  Turns out only a few people heeded the warning and acted, sold their beachfront real estate, or started eclectic car companies.

This story is not for those who will not listen. It is for the small number who will.

What you, as a CIO, already know is you have a problem. Your legacy systems are getting more and more difficult to maintain. Changes are taking longer to implement; new faces are taking the place of the few people left with any depth of knowledge of your systems and the backlogs keep growing. It must be evident; the time is coming and you will not have the resources to keep them all running smoothly. Just like a poorly maintained car, breakdowns are inevitable. User dissatisfaction will grow.

In an earlier article, I suggested that even though CIOs could see a day of reckoning coming, they did not think it was imminent. It would be someone else’s problem. But, to paraphrase Yogi Berra, “it may be getting late around here earlier than we think.” The percentage of IT resources tied up in maintenance has been inexorably increasing. Budget increases are not sufficient to hold the line. The proportion of budgets consumed by maintenance went from 80% to 90% in a fraction of the time it took to go from 70% to 80%. Though the miniscule 10% of resources we have free to deploy on new work precludes any genuine innovation, it does leave a little for some smoke and mirrors distraction like machine learning.

However, the margin for error is small and shrinking. The estimate for 2023, is that IT budgets will increase by 4% versus 2022 which saw an average increase of 3%. For the past two years, budget increases have been well below the rate of inflation. In real terms, IT budgets are being cut. This is likely to be the case again in 2024 and 2025. Four years of growing maintenance demands, declining budgets and the steady loss of institutional memory and the knowledge we had of our systems when they were created may just be enough to tip us over the edge. No new stuff will get done and old stuff will break.

As rational and intelligent beings, what do you plan to do? Obviously, most of you will do nothing. Some of you will make cosmetic changes to create the illusion of action. A very small number of you may ask, “What can we do?” How do you prepare for the day of reckoning, the day the demand for maintenance resources outstrips your ability to provide them?

How do you ensure you are best dressed for Armageddon

First and foremost, it means acknowledging the problem. It means doing what you can do to mitigate its impact. It means buying time until the problem is solved and being ready when it is. It means providing a soupçon of comfort. In the worst case, if the problem is not solved, and you took the action above, you will be able to say, “Well, at least we are not as badly off as them.”

Can the problem be solved? Yes. Will it be solved? Yes. How? I don’t know exactly, but I can give you some clues.

A problem solving technique I find useful goes like this: “How do you stop having wars? You don’t know? It’s obvious. “Have peace”. The response that follows the rolling of the eyes, is “How do you have peace?” I don’t know. That is not the point. The point is you are now asking a different question.

In this vein, wow do you reduce your legacy system maintenance burden? You get rid of our legacy systems. How do we get rid of our legacy systems? I think I have an answer, but it is immaterial if I do. At some point, other people will be asking this question, especially if there is money to be made. One of them will build an electric car.

It may appear that I poured cold water on the idea of legacy systems replacement in my earlier article. Not true. I said current legacy system replacement efforts were not even keeping up with the rate of organic growth in application portfolios. I did not say no replacement method could work; I said the ones we were currently using weren’t working. Why? Because it takes longer to replace a system than it did to build it in the first place. What is the problem? Replacing legacy systems takes too long. How do we solve the problem? I can almost hear you shouting, “Find a way to replace them in less time than it did to build them in the first place!”

Now you are asking the right question. “How much faster would our replacement methodology need to be to have an impact?” My assessment, and the target I am working to achieve is an order of magnitude faster. To solve our legacy systems maintenance e burden problem (by replacing them with better designed, easier to maintain, well documented larger more integrated systems), we must replace 10 systems in the time it took to originally build one. “Replace a system that took a year to develop in five weeks? Obviously impossible, no technology exists to do that. What have I been smoking… yadda, yadda.”

No, not impossible. Very possible. Here is one way it might be done. Hire the same number of people who, in total, were employed developing the ten systems and set them to work. The 30% reduction in redundancy that integrating ten systems should achieve would roughly offset the increased program management complexity (see The Mythical Man Month). They should get it done in a year achieving our average of five weeks. “But that will cost ten times as much,” you say. Yes, I agree, but replacing ten systems at the cost to develop one was not what we were discussing. The question was, “Could we replace ten systems in the time it took to develop one?” Yes, we can, is the answer.

The point is the problem can be solved. Some company, accepting the inconvenient truth that they need to do this will decide it is worth paying what it takes to replace large numbers of their legacy systems. The real global warning is that you may have no choice but to greatly increase your IT budgets. The competition for people will very likely heat up because of the number of enterprises with no alternative but to adopt this solution.

Let’s ask a new question, the one for which you really want an answer. “Could we develop ten systems for what it cost to develop one?” Theoretically, yes. For example, if you pay everyone involved one tenth as much. Outsourcing writ large. “How else might it be done?” Some of you may have noticed that for the past few months, I have been hosting and posting transcripts of weekly collaborative data modeling sessions. Readership of those transcripts is increasing by about 1,000 per week. Perhaps, there is a clue there. 

Nothing is more certain than a solution will be found. The smart money will be on AI, but the answer may be legions of trained cats with mouse pads. It does not matter; a solution will be found. If I am working on a solution, I assume others are as well. Someone will crack the problem. 

Here is the important bit. Taking advantage of the solution will require you to understand all your legacy data. Even if all your legacy applications disappear overnight, you will still need your data. The data is your business. Whatever replaces your legacy systems will need that data. Your customers, their history, your financial data— everything about your business is in the databases of those legacy systems. Your legacy systems are a window into that landscape. The landscape will not change if you replace the window.

Imagine the solution is available and you can’t take advantage of it. One CIO responding to my earlier article said that being a CIO was like being the Captain of the Titanic knowing you had an iceberg in your near future. Now, imagine you were a passenger on the Titanic with that same knowledge. Would you skip lifeboat drill?

Take the 10% that you are wasting on AI and put it to better use. Buy time by reducing the cost and risks of maintenance. Look at all the ways you might do that. Among them are ways to do so that give you the benefit of concurrently restoring your institutional memory. Learn what they are. Learn why your salvation does not lie in data governance, glossaries, catalogs, dictionaries, or lineage. Find a way to way to put total control of your legacy data at your fingertips. Learn what it will take to be best dressed for Armageddon.

Share this post

Rodger Nixon

Rodger Nixon

Rodger Nixon is Data Architect with decades of international experience in enterprise data architecture. Rodger has been awarded several software patents including one covering the use of expert systems and natural language to perform the task of human application developers. The resultant product Gartner Group called “the first 5th generation language”. VP Data Architecture for two major financial institutions and consultant to numerous others. It is the future he saw coming for these enterprises that led him to develop the Open Data Model, SaaS software designed to reduce the costs and risks of software maintenance. In Rodger’s opinion, avoiding a future dedicated to the maintenance of crumbling legacy systems requires a paradigm shift in the way we develop software, one that is data driven. The Open Data Model is his vision of what the foundation for that change might look like and the project that now occupies his time.