Before establishing its integration competency center (ICC) in 2001, a leading consumer electronics retailer was
spending an average of $30,000 per data integration interface, with between five and 10 interfaces per integration project. After the company established an ICC, that figure dropped to $5,000, and eventually $1,000.
Heading up the retailer's ICC at the time was John Schmidt, who likened it to Henry Ford's Model-T assembly line nearly 100 years earlier. But instead of assembling car parts, Schmidt's ICC was assembling integration interfaces.
"It was literally a cookie-cutter assembly process," said Schmidt, now vice president of Informatica Corp.'s global ICC practice.
An ICC, or "integration center of excellence" as it's sometimes called, is a shared services group within an organization whose sole purpose is to streamline and improve data integration processes. By creating and reusing standard integration interfaces -- including extract, transform and load (ETL) and other tools -- and retaining integration-related knowledge and best practices, ICCs can potentially lower integration costs, complete integration projects faster, and help create an IT infrastructure that's flexible enough to accommodate future integration needs.
Companies integrate data for any number of reasons -- for business intelligence purposes, for example, or to meet compliance requirements. Data integration should be viewed as an ongoing process, Schmidt said, not an isolated event. Rather than approaching each integration project independently and starting from scratch, as a systems integrator might, ICCs aim to make integration a core competency.
"Integration is a very repeatable pattern. It doesn't need to be hand-coded. It's the same patterns over and over and over again," Schmidt said. "There's so much reuse that can be leveraged by having a common group do it rather than everybody building their own point-to-point interfaces."
Schmidt said there are common patterns for ETL processes, request-reply mechanisms, and message-based integration functions that can be saved and reused "over and over again."
ICCs also ensure that integration best practices developed in one department are shared with the rest of the organization, eliminating redundancy when new integration projects arise. "It's trying to create a collaborative environment where one might not have existed before across IT silos," said Rob Karel, an analyst with Cambridge, Mass.-based Forrester Research.
Karel said ICCs are still an emerging concept, but they are starting to gain traction at more companies.
ICCs vary in size but share the goal of centralized data integration
Informatica is one of a handful of data integration vendors with a formal ICC practice. It offers customers training and best practices in the form of manuals, books, webinars and onsite consulting to help companies design, establish and manage ICCs. IBM's data integration practice, though not called an ICC, is probably similar, Karel said.
ICCs can take different forms, Informatica's Schmidt said, ranging in size from just one IT worker to several hundred members throughout an organization's lines of business. On average, though, the ICC staff constitutes between 2% and 5% of an organization's IT team, he said. The ICC director is often a vice president, reporting to the CIO, CTO or the director of application development.
At the consumer electronics retailer, Schmidt's ICC, which started with a staff of 13 in 2001 and mushroomed to 180 employees by 2004, was responsible for all integration work, from start to finish. "Whether it was ETL mapping or some kind of message-based interface -- whatever was needed -- we would design it, build it, deploy it and support it in production at some very dramatic cost savings to the organizations," Schmidt said.
Gary Reicher, the vice president of database and architecture services at T. Rowe Price, helped the Baltimore-based investment management firm establish its own ICC in 2003. At the time, the company had just begun building a customer consolidated warehouse with Informatica-based technology.
"We decided at that time that we wanted to build [the customer warehouse] in a way that would be more of a central hub and then allow developers, any of our business lines, to use it to build integration systems," Reicher said. With Informatica consulting, T. Rowe Price managed to get its ICC up and running in less than six months.
Today, T. Rowe Price's ICC, which has six full-time employees, handles integration projects throughout the company, including not just customer data but financial and human resources data as well. It helps the company with change management projects, data migrations and internal audits, Reicher said, and it has saved the company infrastructure costs thanks to a centrally maintained and managed integration effort.
Reicher warned, however, that companies considering launching an ICC should understand the implications for software upgrades. "With an ICC, you run into a little bit of a challenge when you do upgrades, because you're doing it centrally," Reicher said. Not all departments may want to upgrade to the data integration vendor's latest software version at the same time, he said, so more planning is required for software upgrades when an ICC is involved.
Business input critical to ICC success
Forrester's Karel stressed that ICCs -- to be effective -- should include, or at least consult with, not just IT workers with necessary technical expertise but also business stakeholders, in order to define and execute data quality and data governance best practices. In fact, Karel said, he views an ICC first and foremost as a "tactical data governance program."
"Too often, the business provides the requirements and then doesn't see a solution until it's delivered, and the quality component is often ignored," he said. "The quality component is an intricate process. You have to define it, assess it and mitigate it. [Getting business input ensures the ICC is] focusing not only on data movement but also on data quality."
Companies that have a decentralized management structure may also find it difficult to operate an ICC, whose whole point is to centrally manage integration projects. Again, Karel said, executive buy-in and input from the business side is critical.
"If you're considering [an ICC], make sure it has legs. Make sure it has the necessary support and priority to actually make change," Karel said. "If it's just yet another meeting for people to go to, but you don't actually have anything ensuring [that] results are going to come out of it in terms of best practices, policies or improved collaboration, then don't bother."
Does your organization have an integration competency center (ICC) to handle data integration projects? Email SearchDataManagement.com editors with your story!