Creating memories

Software AG-acquired Terracotta is making big moves in Big Data; its in-memory tools have seen 2.5 million deployments. Chief evangelist Gagan Mehra talks to CNME about the hype surrounding Big Data and the company’s love/hate relationship with SAP’s in-memory technology, HANA.

What does your role as chief evangelist entail, and what is the market view from your point of view?
My role is to go out and talk to customers and share our products with them. I discuss with them how to use our products in the most effective fashion. Also, I write a lot of blogs and speak at a lot of events and conferences. There’s a lot of travel involved — it’s a global role.

The market is fairly crazy at this point; that’s why I’m travelling a lot. There’s a big demand on Big Data solutions, though some people are confused. People ask me a lot if they need a Big Data solution — there’s a lot of opportunity, and that’s good. I tell them not to be afraid.

What does Terracotta provide?
We’re in in-memory, so we move data into memory to improve performance. Companies are using this to reduce costs and grow revenue. For example, one of our customers is in travel — one of the largest that deals with flight reservations. Obviously, for them, performance is important — they want to serve as many customers as they can per hour. The more transactions per hour, the better, of course. With their data in main memory, they could start dealing with more requests and more reservations, therefore confirming more transactions per hour.

The thing with us is that we can resolve challenges by allowing hundreds of terabytes of data to be stored in-memory while always guaranteeing performance. If you can focus on business and not data, that’s a good thing.

You have some big players dealing with in-memory technology right now. How do you differentiate yourselves from companies like SAP and its HANA technology?
[Laughs] HANA comes up in every conversation we have. Thanks to SAP, most customers now know what in-memory is. But HANA is a database, and we’re not. We work on any commodity server at the application layer. We don’t need anything specific to run our product — it’s purely software.

It’s a Java product, so if you have a Java problem, we can insert it into your application directly. For this reason, our deployments only take between six and eight weeks. HANA deployments take around six months.

Cost is also a big factor, too. It’s always more expensive to have HANA. Moreover, SAP rarely has customers that are using HANA outside of SAP.

The only good news for HANA is that it does work beautifully with Warehouse — it runs Warehouse much faster.

Regarding start-ups, though, if they can analyse their data in real-time then that will be key. Some start-ups have in-memory already built in directly, because they know they’re going to need it — nearly everyone is going to be using it.

Are the major vendors aware of this, and how will this affect you?
Of course. Microsoft recently announced something in this area, no doubt Oracle is working on their own solution, too, and I believe Gartner said this will be a main trend in the coming years.

But what we do is at the application level. We allow customers to integrate quickly and improve performance — Microsoft and Oracle would take a lot longer. It’s not something we’ll ever end up competing in.