Strata Conference 2014 kicked off today in Santa Clara, California. The event aims to “make the database more accessible to users“, bringing together the biggest names in the analytics industry for a three day gig. Webcasting during the event, theCUBE co-hosts, John Furrier and Dave Vellante persuade Monte Zweben, CEO of Splice Machine, to comment on their recent $15 million series B financing round.

“It feels great. Our company is now on its second stage of financing, growing significantly. Everybody who is working with SQL and Hadoop is so focused on Analytics and having a repository for data science, whereas Splice Machine is really looking to power real-time applications with real transactions. We feel like being the only real-time transactional SQL and Hadoop database, so we got a unique position in the market,” stated Zweben.

Furrier mentioned Hadapt, a start-up theCUBE covered two years ago at Strata, when they won the “start-up of the year” award. “How do you see Splice Machine differing from them?” he asked.”What do you think the market needs and what do you do differently?”

“There’s so much interest on SQL and Hadoop and Hadapt was one of the first organizations to do that. We’re very different from ALL the other players because most of them have developed architectures for analytics only. That means they can’t serve real-time applications. They can’t change records on the fly. They can’t do transactions,” explained Zweben.

“If you are going to be a general purpose database that’s going to power applications on the Hadoop infrastructure, you need all of the services that traditional databases have. Databases like MySQL, Postgres, Oracle, SQL server are the general purpose databases that today are hitting the wall. People are now looking for scale-out solutions. They can scale-out on NoSql, but they give up on all the SQL services. They can scale-out on the new SQL infrastructures out there, but they are unproven and proprietary architecture. OR, they can scale-out on Hadoop, which is proven technology at petabyte and beyond scale. That’s what we’re trying to bring people to, powering real-time applications.”

“Splice machine is fully ACID,” added the Splice Machine CEO. “The question is, what use-cases is that good for?”

Zweben gave a very elaborate explanation: “Many people think that ACID properties are just for traditional transactional types of operational applications – powering a website, doing financial transactions, e-commerce – but transactions are required even for OLAP (Online Analytical Process). If you want to update secondary indexes in a SQL database, at the same time as the data and keep that consistent, you need a transactional context. That requires ACID propperties, a traditional architecture that allows you to keep those atomic, consistent and isolated transactions in the database,” he said.

“Even if you are doing a large-scale reporting application or an analytics application, you do need transactions and ACID properties. And that’s why we do what we do,” stated Zweben.

Common Misconceptions

A lot of people still perceive Hadoop as a static data repository, as something that you do data science on. But Monte Zweben is confident that “we’re going to start to see real-time SQL-based applications powered by systems like Splice Machine.”

He relayed the use-case of a customer, a direct marketing services provider to large-scale retailers who has implemented an off-the-shelf campaign management software package powered by Oracle. When they hit the wall they turned to Splice Machine for help in scaling out their business to make it more profitable and to allow growth. Splice Machine proved to perform better on much less expensive hardware. “We’re proving that Hadoop can power a real-time application,” said Zweben.

Asked what was the catalyst for change,  Zweben commented: “I don’t know why exactly they chose us, but I know some of their pain points: they had performance problems in the Oracle architecture and they wanted to expand. They were looking on staying with Oracle and solving their issues, but that was quite expensive.”

Metrics of success

“I’d like to go to 2015 with thousands of nodes deployed, south of $ 5million contracting activity for the first year of a start-up company having a generally available product,” confessed Zweben.

Splice Machine is now in private beta, with a headcount of 30 people. According to their CEO, sometimes soon they will be in public beta.

“The interest level we’re getting tells me this is a market that is quite hot at the moment,” concluded Zweben.

Check out all the segments from this week’s #BigDataSV live stream here.