Adrian Bowles: You’re not seeing a big risk in having to swap out something that you’re comfortable with something else just for the IoT?

Mike Piech: Right. There certainly will be new and more specialized, I would say, incremental pieces of middleware and components and maybe even types of editing tools that one might use. It’s exactly that, it’s incremental. It’s not a rip and replace type of proposition.

Part of making the volume tractable is filtering and selecting, aggregating, et cetera. For that aspect there are tools like business rules, management systems, so for example, in our middleware we provide one of those. On the variety front there are also some bits of middleware that can be very helpful. For example, there’s data virtualization which is a layer of software that would allow data
sources that have different formats, different structures, and so on, to be aggregated in a middle man virtual layer where those different formats, those different structures can be normalized and then presented to the consuming layer as if all that data was [00:03:00] in one single, nice, clean, unified database, data structure, et cetera.

That’s a bit of help on the variety side. When you get to the velocity aspect, one of the key technical considerations there is to do as much as possible in memory. That’s where an in memory data grid really comes into play where you’re not having to write to disc and then consume from disc or have too many high latency interim steps and processing of the data. If you can everything in memory you can get a lot closer to real time, handle a lot greater velocity and Page 2 of 2 that’s where, again, a product, a middleware solution such as in memory data grid can really come to bear.

Adrian Bowles: Sounds like the key thing to remember is that if you’ve got a good middleware stack to begin with people shouldn’t sweat it?

Mike Piech: Exactly. That’s absolutely the idea.

Adrian Bowles: Great, thanks for joining us today.

Mike Piech: Thank you Adrian.

Adrian Bowles: Take care.