This article is more than 1 year old

Ex-Google & Facebook bods are on a Quest for the Enterprise

Big Data compute meets Java culture with Metanautix

Number-crunching without the overheads

Quest, according to Vassilakis, takes the SQL out for RDBMS, letting you do the analytics without the need for relational plumbing.

“I felt there was a huge opportunity in analytics; start ups are all about timing, and the enterprise thing was one of the big elements,” Vassilakis said, but it's still “tough to leave when you’ve been in a company like Google for eight years".

Google used Tableau and Hyperion with home-grown tools to probe Dremel, and Quest uses Tableau, too. “Part of our idea, and why we are focused on enterprises, is we assume you have a BI analytics tool and we integrate with it, and we also integrate with your storage,” Vassilakis said.

We’re talking Excel-style macro jockeying for big data and corporate data.

Dremel is the inspiration for Quest, built separately at Metanautix, Vassilakis stresses. So what’s different?

First, scale: Its minimum cut off is no longer thousands of servers; it can run on a laptop.

Second, diversity: Quest works on other people’s servers, not just Google's – AWS, Rackspace and vCloud Air - while it also works with VMware on prem.

It also eats legacy data by chomping on standard SQL in Teradata, Oracle and SQL Server plus Json in MongoDB. Metanautix has written adapters to things like MongoDB, while also re-using adapters, such as ODLBC drivers.

And there's Java: To say Java is “big” in this market would be an understatement. Quest is written in C++ but Metanautix has embedded a JVM into the code.

The JVM means Quest can use Java and Jython to create user-defined functions for integration with that legacy enterprise code that is often written in Java. The marriage wasn’t easy and language aficionados of both camps will be offended.

What’s next for Metanautix and Quest?

“Technically, it was challenging,” Vassilakis said. “It takes a lot of hard, low-level programming and resource management if you want to do it well and reliably."

“That was part of the value proposition here – there’s a class of guys here who can do that. Some people are a little bit offended but they usually see the value of it," he added.

Another big difference is image and flow processing. Google used Dremel to query tables but not images; that was seen as YouTube’s job.

The goal is to turn Quest into something that can process structured-with-unstructured data – so more work on support for photo and videos. The focus right now is work on Tableau on the front-end and Teradata and MongoDB at the back, valuable niches that are not well served.

The natural next extension is integration with SAP and Salesforce. Working with open-source, ODBC, and Java Metanautix hopes to attract SI partners writing adapters and plug-ins.

After that? Well, “long term, the number of applications understanding data better like machine learning and fuzzy joins,” Vassilakis said. ®

More about

TIP US OFF

Send us news


Other stories you might like