Top Ten Best Practices for Data Integration
Use these guidelines to help you achieve more modern, high-value and diverse uses of DI tools and techniques.
If integration were a pizza, how much would a large cost at your company? What about a medium - any idea? Can you at least price a small, simple integration cost?
It may sound silly to compare integration pricing to pizza, but it's an approach that helped Best Buy reduce the average costs per integration point from $10,000 to $1,000 within three years, says John Schmidt, who lead this cost-reducing effort at Best Buy years ago.
Schmidt is now the vice president of Global Integration Services and Lean Integration Practice Leader at Informatica, but he's also well known as an advocate for Integration Competency Centers and co-author of "Lean Integration: An Integration Factory Approach to Business Agility." Recently, I spoke with Schmidt about his idea of integration point analysis - a fancy term for measuring the costs of integration points.
"You contract some work to Accenture, some to IBM and some others to PricewaterhouseCoopers. One of them quotes you $50,000 for the integration work, another is $100,000, another is $150,000 - which one is right?" Schmidt said. "Is the $150,000 more expensive or are they actually more efficient but the reason it costs more is because of higher complexity? It's hard to get any kind of apples-to-apples comparison until you can define the deliverables in some kind of standard unit."
Knowing integration point costs can also be a tool for driving efficiencies and therefore reducing the costs of integration over the long term because it gives you a way to measure work, or output, he explained.
"You want to be able to say, 'We're going to define a standard integration point and we'll put some definition on what we mean by standard and we're going to agree that we will pay you, the supplier, $10,000 per standard integration point. Year over year for the next seven years on this contract, we want to see a 10 percent reduction in that per-unit cost,'" Schmidt said.
This isn't a new idea. Functional point analysis has long been a method for calculating the functionality or capability the business is getting from software. But when applied to integration, it becomes extremely complex, and so few organizations adopted that approach, he said. His recommendation calls for a simpler method of comparison: Look at large, medium and small integrations. It's an easy system that's simple for business leaders to understand. The trick is defining what each size means at your organization and attaching costs to those.
It doesn't have to be overly complicated, though. Schmidt says an organization should be able to establish a baseline for integration point costs in a short amount of time - say, weeks or a few months.
"It's a combination of analyzing whatever data they have, plus interviewing staff and piecing a picture together. That gives you a preliminary rough baseline to say, 'Okay, this our baseline cost per integration point,'" he said. "Then, from there, putting in place processes to capture data as you move forward so that as developers, analysts and team members are working on it, they have a way to capture effort being spent on integration points. Over time refine your estimates."
Schmidt explains in more detail why it's important to come up with an integration point analysis and how it can impact costs in a recent IT Business Edge Q&A. You can also read his original Informatica blog post on the topic, which focuses more on why his approach should replace functional point analysis.