The ability to embed analytics within transactions in real time can be traced back to the mainframe platforms that have enabled this capability for decades. But as data is increasingly processed and analyzed at the point where it is created and consumed, the need to analyze transactions and other events in real time has become more pressing.

VentureBeat caught up with Hazelcast CEO Kelly Herrell to get a better appreciation for how in-memory data grids and event streaming processing platforms are being combined to drive the new class of translytics applications that are transforming IT.

This interview has been edited for clarity and brevity.

Venture Beat: Translytics as a concept has been around for some time now. What’s actually changing?

Kelly Herrell: The big move is data in motion. IDC most recently said we will generate more data in three years than we did in the last 30. That’s going to be a head-snapper. The preponderance of data is newly generated. It’s an event or it’s a stream. It could be an event from Kafka. It could be an IoT stream. It could be a stock trade. It could be a clickstream. In this new world, there’s an opportunity to process data in real time rather than wait until it’s stored. It’s about doing analysis and performing transactions on that data in the instant in which it is born and then combining it with context from databases into a single unified workspace.

VentureBeat: Where does that fit in the current IT environment?

Herrell: This is basically a new layer in the architecture that is rapidly evolving. It loves databases and systems of records because that’s where it pulls its context from. It also loves sources of data in motion because that’s where it pulls its events and streams from. It unifies those two things. It’s extremely complementary. What you’re seeing is customers building a new breed of applications that are strategically incredibly valuable.

VentureBeat: Can you cite an example?

Herrell: We’ve got customers using our two products, one is an in-memory data grid and the other is an event streaming platform in conjunction with each other to do risk calculations for large financial services companies. They conduct all these trades all day and then at the end of the day they stand up a batch process to calculate how much risk they’re sitting on. That batch process runs all night long. In the morning, they need to make sure they have enough cash to cover those risks. That’s an overnight batch process based on old, stale data. We’ve got customers turning that into an all-day continuous process so they don’t have to wait until they wake up in the morning to see where they stand. At the end of the day, they know where they stand.

VentureBeat: Where does that platform typically run?

Herrell: It’s going everywhere. This is happening in the cloud. It’s happening in the datacenter, and it’s happening at the edge. Our system is very small. It’s a 12MB JAR file on a single node. We do demos on Raspberry Pi systems or as big as a massive cluster that we spun up and Amazon to process 1 million events per second, with 20 milliseconds of latency for events. We’re also working with a large streaming company to turn that billion into 10 billion events per second. Our mission is to empower users to act instantaneously on data everywhere. Not just look at it but actually act on it.

VentureBeat: Where does AI fit in that equation?

Herrell: AI is simply going to be a part of applications. It’s just going to be another set of logic that goes into applications. There is not going to be an AI industry. It’s just an additional way to write very intelligent applications. And that those intelligent applications can perform in real time with very low latency. That’s where you start unbuckling the value.

VentureBeat: What’s the biggest challenge getting organizations to appreciate this new IT architecture?

Herrell: I think the biggest thing is people kind of standing back and saying, “Wait a minute, what is this big-picture change that’s happening?” Every piece of data is born in real time. This is going to be a new category. It doesn’t replace anything. It simply takes advantage of [the data’s] existence and pulls pieces together in a unified way to create a new breed of applications.