Data seems to be a forgotten component of a good system architecture. In cloud migrations, I’m often taken back how the data that was bad on-premises is worse in the cloud.
What does bad data look like? Poor data structures, redundant data, no single source of truth, and security and compliance issues all top the list. Indeed, too many enterprises have just kicked the data can down the road using cloud computing camouflage.
Here are two core data recommendations for properly migrating to the cloud.
1. Create a common data model that spans all enterprise data
This common data model can be a virtual data model, such as by using data virtualization software. Or, this can be just a model that’s conceptual or logical that IT uses as a reference. What’s important is that the model reflects all the data and data entities that the enterprise has to deal with, such as the notions of customer, product, inventory, and sales.
You use this model as a common reference point for all enterprise databases, whether cloud or om-premised. That gives you a focus on a single source of truth, and it gives you an atomic level understanding of all data, data elements, and their meaning. This model can be further tied into a master data management system or into data-governance systems.
2. Fix your data as it moves to the cloud
You need a few extra steps in the migration process, including changing the structure and perhaps even changing the databases. Of course, the applications have to change as well to use the new and improved databases. (By the way, those new and improved databases use the common data model from my first recommendation.)
Although this seems like a no-brainer, the reality is that many enterprises skip the data improvements or repairs during migration because of the cost and the time delays. They assume they will get to the data in the cloud at some point, and then fix data issues there. But that “fix it later” approach is typically more problematic; it’s like changing tires on a moving car.
Data is a fundamental part of any system, and most enterprises need to improve their data before it moves to the cloud. I hope you’re the cloud migration team that get serious about fixing data and increase the value of your cloud migration fivefold as a result.
