Recently I sat down with Massimo Pezzini, vice president and fellow at Gartner, to discuss event processing. Pezzini specializes in in-memory computing, so—not surprisingly—the first question we came to is how these two technologies relate.
In-memory computing describes technology that stores data in memory (rather than on a disk), usually for fast initial or repeated access. Event processing is a system that responds to incoming events, usually in an automated asynchronous fashion, in contrast to systems that wait for scheduled or human-initiated tasks. The two ideas are certainly not in conflict. In fact, once you’ve put them together, it’s hard to imagine an event processing system built any other way than in-memory, and why you would want an in-memory computing system that wasn’t kept up to date with events.
In-memory computing is often used for operational analytics, resource allocation, fraud detection, or best offer selection. Event processing accelerates these use cases with live updates to data, and with automated action and reaction as circumstances change.
In-Memory Gives You the Right Now
In-memory systems are prized for their ability to process data quickly, and, in particular, to process the same data repeatedly—analyzing it in different ways or updating it as circumstances change. The value of not having to wait for a report to run against a traditional warehouse is being able to make operational decisions with up-to-date information and to trying out different ways of analyzing data.
In-Memory Is Better When It Is Live
In-memory systems are most attractive to users not only when the data analysis is timely and interactive, but when the data itself is fresh and relevant. This means that as operational realities like inventory, users on site, and staff locations change, information is immediately reflected in the analytic systems. Fast Data is more complete and better able to drive problem solving. Operators can see that when they take action to resolve a problem—manually, automatically, or outside of the system—the resolution is immediately reflected in their view of the data.
Making In-Memory Computing Automated and Active
Once you have an in-memory computing environment, you are going to want to automate action based on it. As new data comes in, analysis can be updated, opportunities detected, and operators alerted. But oftentimes the first response to a new opportunity or threat can be an automated one, freeing operators to focus on more complex problems. Event processing is how you take a live in-memory environment and make it active, enabling automation.
