Energizing the elephant

ScaleOut hServer V2 adds its own shot of real-time analytics functionality to Hadoop MapReduce

Lucy Carey

New version promises enhanced speed and flexibility for Hadoop adherents.


In well under a decade, Hadoop has become the go-to solution for
big data analytics, offering scalability, fault tolerance, and
critically, mass accessibility. But it’s not a fix all – there are
myriad uses and functionalities within the colossus of the
framework yet to be unpacked. An increasing number of users are
investigating software with the real-time analytics options which
fresh out of the box Hadoop simply cannot currently offer, such as
Storm. Yet, as the launch
of hServer V2 by ScaleOut Software this week demonstrates, there’s
a host of solutions out there ensuring Hadoop remains a force to be
reckoned with.

ScaleOut Software’s hServer, an in-memory data
grid for Hadoop MapReduce that first launched earlier this summer,
works b y bypassing the batch-based framework and enabling Hadoop
to work in a real-time manner, acting as the catalyst for a whole
new wave of implementations.

With the addition of ScaleOut Software’s
MapReduce engine, the company reckons
that applications can run 20 times faster than with plain old
Hadoop. It’s probably not ideal for titanic sized batches of data,
but should be quite effective for analyzing real-time operational
data, as well as testing and debugging small segments of full
MapReduce workloads.

With Hadoop’s lack of speed a bugbear for many
vendors, anyone who can conceivably offer a way to make the
plodding pachyderm run a little swifter is guaranteed to make a
profit – meaning there’s a raft of similar options out there. When
we asked ScaleOut Software’s CEO Bill Bain what differentiates
hServer V2 from the rest of the herd, he commented that, ScaleOut
Software, “Have directly leveraged our deep in-memory parallel
computing technology and customer experience to introduce an
in-memory Hadoop MapReduce execution engine that is fully
integrated with our in-memory data grid”.

According to Bain, this provides “a computing environment
to run Hadoop MapReduce code on [with]
fast-changing data sets”, and enables continuous analysis on
live operational data with Hadoop MapReduce. He added that, “While
there are other products that accelerate various aspects of Hadoop,
our focus on in-memory computing puts us ahead of the curve in
applying Hadoop to meet real-time needs”.

comments powered by Disqus