Ex-Google & Facebook bods are on a Quest for the Enterprise
Big Data compute meets Java culture with Metanautix
“Enterprise” and “Java” aren’t things you’d expect from Google — and certainly not Facebook.
Talk Google and you think search, ads and YouTube. Your second thought might go to Google Apps and Chrome.
Say “Facebook” and what do you think of? A firm making fistfuls of money from ad men mining its members’ data.
Their diaspora are different, as two former employees have come together to follow a common dream.
Theo Vassilakis, a former Google engineering director who led on query engines, visualisation systems and data generation pipelines, and Toli Lerios, a former Facebook senior software engineer who worked on image and video algorithms, pushed the eject button from their respective motherships recently.
Their new technology offering is called Quest, something they call a data compute engine from their firm Metanautix, which was founded in 2012 with $7m backing from Sequoia Capital.
The idea is that using industry standard SQL you can search a multitude of big unstructured, structured and relational data sources through one engine. Metanautix Tuesday announced a free, single-user edition of Quest.
As happens with such stories with a Google connection, Quest’s roots lie in a Google project – an ad-hoc query platform called Dremel, which, in typical Google style, scales to thousands of CPUs and petabytes of data. Vassilakis helped lead development of Dremel.
Dremel, being a Google baby, worked for Google, eating a non-standard variant of SQL. It only worked on Google systems.
“Dremel was not easy to run,” Vassilakis told The Register. The idea with Quest, however, is to take the lessons learned with Dremel to enterprise.
An epiphany, apparently, gained working with non-engineering types at Google, helped Vassilakis.
During his time at Google Vassilakis inherited 75-80 engineers, with part of their role being supporting finance, sales and ops. It resulted in a clash of cultures, with the IT types not overly keen to get button-holed into becoming the big data version of macro jockeys by writing code for their analytics and query jobs.
“We had long periods of spirited disagreements and then we, at a personal level, figured out why it was the way we were,” Vassilakis said.
“That setup creates friction, especially since the business requests are usually legitimate, very important, and often go straight to decision-making executives. Part of what we did inside of Google, and what was the inspiration for some of Metanautix, is to try to cleanly separate the part that IT and engineering does, and the part of the work that the business can do for itself.”