Warning Signs of Bad IT Architecture
Chances are someone at some point spent countless brain cycles planning your organization’s IT architecture before handing the grand plan off to someone else to build it out, and then to someone else to maintain it as your computing environment inevitably grew. And, chances also are, somewhere along the line, best intentions faded in the face of expediency, departmental politics, and general mismanagement, eroding what was once a coherent architecture management strategy into an ongoing series of independent, case-by-base decisions about each technical component.
How do you know if your organization has strayed from the path? Here are nine warning signs that bad IT architecture has taken hold of your organization.
Manual re-keying might not be the biggest cost companies pay from bad architecture, but it’s certainly the most obvious one. Hiring human beings to serve as the interface engine connecting incompatible applications isn’t just expensive; it’s de-humanizing.
Collection of Point Solutions
Everyone wants their work supported by a “best of breed” solution. Define “their work” too narrowly, though, and everyone has to visit so many applications to get their work done that there isn’t enough time to get their work done.
Meanwhile, unless IT spends a lot of time building interfaces to connect all of these point solutions, you’re back to re-keying again.
Every business application solves business problems. Solving business problems is good, so solving them more than once must be even better, right?
Of course not, and yet a lot of companies keep lots of redundant applications around, either because they overlap but still have a few unique areas they support, or because they’ve grown through mergers and acquisitions but aren’t very good at integrating everyone into one business after the papers have been signed.
Either way, the money spent to support all of this redundancy is pure waste.
Very often, different applications need the same information to get their jobs done. You have two choices: Point them all to the same underlying database, which isn’t always possible, or synchronize their separate databases, which is often pretty messy.
Or there’s always that manual re-keying option....
Too Many Interfaces
When you have redundant data and you decide to keep it synchronized, you need to build an interface. Even if you don’t, you often have to feed one system with results from a different one.
Either way, the more systems and databases you have, the more interfaces you end up building. It’s better than not having them, but as they accumulate, your architecture becomes more and more fragile, and you spend more and more time managing the interfaces instead of building new functionality.
So you decide to solve your interface dilemma with an elegant enterprise application integration system, or a services bus, or some other form of middleware-plus-metadata that keeps everything clean.
And then, your developers figure two things out: (1) what your cool new system does is make solving the easy problems even easier; and (2) it doesn’t solve the hard problems at all. So instead of arguing with you, they rebuild the same old spiderweb of interfaces, but hide it inside the EAI system so you don’t know about it.
Kludges and Workarounds
Maybe you were competing with an outside developer who lowballed a project. Maybe the business sponsor insisted on too short a deadline. Or maybe building a solution well would have ruined the business case for the project.
Whatever the reason, you wake up one day to discover a lot of your systems are held together with Band-Aids, chewing gum, and duct tape.
If you’re lucky, nobody will notice until after you leave or retire.
It’s mission-critical! It satisfies the business need perfectly! What do you mean you have to spend money to maintain it?
When you’ve built something on a version of Visual Basic that Microsoft hasn’t supported in a decade, that can’t read and write from any version of SQL Server that isn’t at least seven years old, and the only versions of Windows they’ll run on don’t have drivers for any of the printers you have in production -- that’s what you mean. You have to spend money to maintain it.
You see a bunch of warning signs. You organize an enterprise technical architecture management group. You hire an expert or two. And their productivity is enormous.
Enormous, that is, if you measure productivity in terms of the number of white papers they publish. Changing how work gets done in IT? Of course they’ll change it. So long, that is, as everyone reads their white papers, admires their business, and follows their instructions.