SQL gets supercharged with C++

Facebook alumni launch ‘world’s fastest database’ with backing from Ashton Kutcher


A couple of ex-Facebook engineers have just launched MemSQL, an in-memory database system which claims to be “30x times faster than conventional databases on disk”

Ex-Facebook engineers Eric Frenkiel and Nikita Shamgunov
today announced the
launch of MemSQL, an in-memory database system claiming to be “30x
times faster than conventional databases on disk”. Utilising
familiar elements such as SQL-style queries and the standard MySQL
API, MemSQL boosts performance by storing data in main memory and
processing it in parallel, much like some NoSQL solutions.

The startup certainly isn’t shying away from the limelight,
claiming in bold text on
the front page of its site that MemSQL is “THE WORLD’S FASTEST
DATABASE”. The hype has already impressed a group of institutional
and individual investors — including Dude, Where’s My
 star Ashton Kutcher and venture firms First Round
Capital and New Enterprise Associates — who furnished MemSQL with
$2.1m in seed capital back in March 2011. The company has raised a
total of $5m to date.

 to GigaOm, Frenkiel compares it to HipHop,
Facebook’s internal (but open source) PHP compiler, which converts
the scripting language into optimised C++. We’ve definitely heard
good things about HipHop — during a session at IPC Spring in
Berlin earlier this month, Facebook’s Sara Golemon waxed lyrical
about its efficiency, claiming that it allowed the company to delay
a massive server upgrade and save a significant amount of

While today’s press release is datelined from
San Francisco, MemSQL also maintains an office in New York, and it
looks like it’s focusing on the financial sector for at least some
of its initial customers. “The solution is ideal for applications
that require fast processing of machine data, including financial
services”, claims the company — to top things off, the corporate
website features the Manhattan skyline as a panoramic

We’ll wait for the initial hype to die down before passing
definitive judgement on MemSQL, but it certainly looks like an
interesting initial product, sitting comfortably in the niche
between MySQL: familiarity and NoSQL performance and scalability.
One to keep an eye on . . .

(via GigaOm)

comments powered by Disqus