Recently, I interviewed David Lyle and John G. Schmidt, authors of a new book, "Lean Integration: An Integration Factory Approach to Business Agility." Both men work for Informatica, and Schmidt had first written about the idea of applying the seven lean principles of manufacturing to integration on Informatica's site. I thought it was an intriguing idea then, but I had no idea how effective it would actually be.
And neither, I think, did Lyle or Schmidt. Certainly, they knew the principles would work, but after compiling case studies for their book, I suspect even they were surprised at how well and repeatedly lean practices improved productivity and delivery speed for delivering integration.
"This isn't just theory, this actually does work and we've seen it consistently, when you apply this waste-elimination kind of mindset and continuous improvements, you should have about 90 percent cycle-time reduction, which is the time it takes from something requested until it was delivered, and about 50 percent productivity improvement on average," Schmidt said. "We've got a number of different cases to support it, so we're pretty confident in those numbers."
If you're not familiar with lean's seven core principles, Joe McKendrick recently wrote a piece describing them.
In brief, they are:
The book walks you through these principles and how you can apply them to IT integration. It's probably most useful for really large organizations, with what Schmidt and Lyle call "integration factories," but I think there's a take-away here for small shops, too. In fact, in part II of my interview, we discussed whether you could apply the principles to hand-coding. The consensus seemed to be that you'd be better off moving away from hand-coding, but if you're going to be stubborn about it, at least reach some consensus on standard practices.
One of the things we also discussed was how you apply step four-empower the team-to integration. I think this point is especially important for managers and executives to consider, because I do think there's a tendency to treat integration as a very technical, tactical process, essentially reducing your developers to line workers on a factory floor. This only contributes to the idea that integration is a project, rather than something that actually can be very strategic.
So I asked Schmidt and Lyle how companies can empower developers. They offered some concrete steps to take to give developers, data analysts, DBAs, software engineers, data architects, business analysts-whoever handles the integration - ownership in the process.
Integration isn't cheap. It only makes sense to do what you can to make it more efficient-and, one would hope, reduce the costs. Even if your results are only half as effective as the companies Lyle and Schmidt studied, lean integration would be worth pursuing, or, at least, investigating.