Join Newsletter
Trusted Business Advisors, Expert Technology Analysts

Taneja Blog

Taneja Blog / Big Data / Data Center Systems

Nothing’s Too Fast for Operational Intelligence - ScaleOut Software’s hServer

There are a lot of HPC technologies coming soon to a data center near you! The latest offering from ScaleOut Software, known for their in-memory data grid solutions, is a customized in-memory data grid for Hadoop. This enables a blistering fast big data style real-time analysis of dynamically changing data.  Solutions that use this are processing live operational data into actionable intelligence - financials, reservation systems, live customer experience, etc. 

While there are a quite a few speed optimization and streaming data packages being developed, hServer is built on proven production quality software. ScaleOut has a lot of field experience and hServer's grid is made to be not only elastically scalable, but also highly available.

As the name implies, this is a scale-out cluster "grid" solution. Unlike HDFS, when you add a node to hServer, it will automatically repartition and rebalance. Technically you can stream data into this grid in real-time and apply map reduce style computations in memory (actually something that their existing analytics solutions did before). In this way, total data motion is minimized. But since the world is Hadooping along, the new hServer brings some java interface libraries that enable Hadoop jobs (map reduce jobs on Hadoop) to acces the in memory grid while minimizing network overhead.  Since data is stored in memory in key-value pairs, I think of this as a type of converged noSQL/big data hardware solution.

In addition to enabling Hadoop to read/write from the in-memory grid, hServer comes with a library so you can quickly set it up as a read cache between Hadoop and HDFS. This won't require re-developing your Hadoop processing but could give you a big speedup. They report an 11x latency reduction over native HDFS reads.

To entice folks further, there is a free community edition for up to four nodes with no restrictions on using it in production, so now everyone can add an in-memory data grid to their big data platforms.  Definitely worth trying out if you want to develop a more operational edge to your computing.

  • Premiered: 04/16/13
  • Author: Mike Matchett
Topic(s): ScaleOut Software Hadoop Big Data HDFS


There are no comments to display. Scroll down to leave your own!


Leave a Comment

You must be logged in to comment. Click here to log in or register if you don't have an account.