Founded in 2003 by Ari Zilka, Terracotta, a wholly owned subsidiary of Software AG, develops breakthrough software that delivers snap-in performance, scale, and availability for enterprise applications. Terracotta’s technology can store up to a terabyte of data in-memory, which can boost application and system performance and scalability significantly. At the time of its acquisition in June 2011, the company had 55 employees located in development centers in San Francisco and New Delhi. The company’s customer base is filled with names like JP Morgan, Adobe, BBC, and Hitachi, among others.
Sramana Mitra: Hi, Ari. Let’s start off with some context about both Terracotta and your personal background.
Ari Zilka: Terracotta is a data management technology provider. We sell direct to enterprises. When I say data management technology, what I mean is play in the space that databases play in, that no-sequel plays in, that Data Grids plays in. You would see us competing often with companies like Oracle or IBM or Tibco in accounts. What is our unique value and our differentiated go-to-market? We offer access to terabytes of data, which is the amount of data typically stored in a large enterprise database. But we keep it in memory, and we keep it in memory at low cost. We’re giving people access to all their critical business data about one thousand times faster than Oracle – well, in the best case – one thousand times faster than a technology like Oracle or IBM’s traditional databases can deliver. On average, we’re a hundred times faster. We take customers from seconds to milliseconds and from milliseconds to microseconds by moving their data in memory through our product called Big Memory.
SM: How is the product or technology deployed? How do you go to market?
AZ: Great question. [Our] model is open source with a commercial backing behind and a traditional go-to-market. What we do is for Java and .Net programmers, we provide one of the world’s most ubiquitous data access frameworks and data management frameworks. It’s embedded in a lot of products that they would download [in] packaged applications from other vendors’ technology. JBoss, Atlassian, and IBM use our technology. SugarCRM, Salesforce, and all kinds of vendors embed us in their products to make their database access go faster and to make their applications scale better. But end users use us directly.
More than a million developers use our products every day. We give away the interface to our data management layer, and then we charge you for the enterprise version of the data management layer.
SM: I see. So, is the entire trick of the solution in the data management layer.
AZ: It is not. It’s a great question because if our data management is just about performance, won’t Oracle catch up to us? If our data management is just about an interface that Java developers use every day because it’s free, can’t someone else give away [his] interface free and compete with us? And the answer to both of those questions is no, and it’s because that’s not the only thing we do.
So, we give away the interface free. We give away a little bit of the data management layer free. But our data management layer is not just fast, it’s also intelligent. It is capable of doing what I call fast and loose data management, which means best efforts; try to write this data down for the business. If you can’t, just keep moving. Don’t wait too long. [It’s] very safe and careful data management, meaning do not let the user think this data was written and then lose the data. The two extreme examples would be a website where you’re trying to comment on a friend’s photo. If that comment doesn’t get submitted, so what? Submit it again. Whereas for a bank, if you are doing a bank-to-bank wire and the account on the outbound side has been debited, you better credit the inbound account before you complete the transaction or the money’s now in limbo somewhere. The two banks have lost, and neither customer has access to it.
Our system has the intelligence to offer fast access to data, careful access to data, large amounts of data and multiple interfaces to that data, not just one that we give away. It can look and feel like a database with rich and robust query and analytics. It can look and feel like a no-sequel store, very cloud friendly where it is just easy to use and in tune with what the Web 2.0 developer wants to see. Because it can change itself like that, it becomes what I call a universal interface to all your data. It’s a very chameleon-like tool. That is at the heart of why people take it down and use it – not because it’s free, and not only because it’s fast and can handle your entire enterprise’s data set and keep it in memory, but because it is adaptable.
It works with the data sets I already have. If I have a legacy system, it’ll work with it. If I want to go to the cloud and be leading edge, it works there, too. Its unique value is in its flexibility.