In my last posting, I listed the functional criteria that data integration platforms must support in order to address the comprehensive needs of most organizations. But it’s not only important to consider what the platform can do—it’s important to think about how things get done. One key aspect is how unified the platform is, as unification can radically simplify deployment and management of the platform.
What does “unified” mean when it comes to a data integration platform? There are a lot of ways to define this, but the most important is from the viewpoint of the users of the platform. You can have an extremely elegant, unified technical architecture underneath a platform, but if users still have a disjointed experience, that architectural elegance doesn’t really matter. A unified user experience is a product of both the design of the user tools, as well as how metadata is shared.
From a tools perspective, a common look and feel, and cross-tool integration, is key to making it easy for users to ramp up on the tools, and to reuse assets across them. On the other hand, a unified experience does not mean trying to cram every single function or capability into a single tool.
That is not practical because there are many different roles involved in data integration—from stewards and analysts to architects and developers—and they each have different tasks to do, and bring different skills to the table. So the tools have to be tailored for each role, but still foster collaboration across the different users and roles.
A key underpinning of this collaboration is shared metadata, and the bar for success in sharing metadata is how the users experience it. Again, each role may need to work with a different type of metadata, or a different view of it. For example, an analyst may need a certain view of metadata, working with business terms and rules. That is different from the technical metadata a developer needs to see in the form of physical data sources, transformations, etc.
But while they may need different views, the underlying metadata should be common. The key is making the right metadata available to the right people at the right time in the right way. It is critical that the data integration platform is designed so that each tool can access the relevant metadata and make it available to the user, regardless of where the metadata physically resides.
Looking across the various data integration tools used in your organization, how unified are they? If different users have a highly disjointed experience, perhaps it’s time to take another look at how you’re approaching data integration in general, and whether adopting a platform approach makes sense.