There's been a lot written in the past three months about moving big data analytics and applications into the cloud. The basic concept makes sense for two key reasons:
- Advanced processes such as artificial intelligence, machine learning and neural networks run on big data, which often resides in the cloud.
- Renting cloud instances is usually less expensive than buying custom hardware to run advanced analytics.
But if you're the CIO of a large enterprise, there's a good chance your in-house data scientists have already developed a bunch of home-grown analytics for dealing with specific challenges that arise in various business units across the enterprise.
If those home-grown apps are popular and effective, your CEO will probably ask you about scaling them for the rest of the enterprise. That's when they can become headaches for IT.
An application or solution that's written for enterprise architecture might not work as expected in the cloud. Enterprise solutions are written with vertical integration in mind. In other words, they have to work smoothly with the software running immediately above and below them.
Cloud-based analytics, on the other hand, usually require a high degree of horizontal integration, because they're constantly adjusting to huge volumes of fresh data from multiple sources.
It takes a different mindset to write a cloud-based app. You can't simply take an app that's designed to run on enterprise hardware and transport it magically into the cloud.
Explaining the differences between on-premise enterprise architecture and cloud-based architecture can be difficult, especially if the audience isn't particularly interested in the technical details and just wants to see quick results. But you can't simply pretend that it's easy to scale a local solution into an enterprise solution by merely shifting it into the cloud.
Most executives don't understand why an app designed to run on local equipment won't necessarily run in the cloud. My advice is to begin talking about scale issues now, so they don't become surprises later on.
Take the time to develop a short presentation or video explaining the differences between enterprise architecture and cloud-based architecture. Here's an example of a good video from Intel on a related topic. Begin the dialogue now, and you will avoid a lot of headaches down the road.