2 Comments

Summary:

As promised, storage kingpin EMC has integrated its Isilon NAS product with Hadoop in a way that will bring Isilon’s OneFS file system to bear on data. EMC isn’t alone. Vendors from Amazon to Oracle are trying to tame this big data beast.

plumscreen Shot 2012-01-31 at 12.05.25 PM

EMC on Tuesday launched its Isilon NAS product melded with EMC’s own Greenplum Hadoop distribution.

This is the latest in a series of vendor moves to address customers’ need to make sense out of massive amounts of structured and unstructured data generated by people, by applications, by servers and by devices. The resulting trove is too much for traditional relational databases and ancillary tools to handle, which is why legacy IT vendors from IBM to Microsoft are diving into Hadoop, the framework seen as best suited for handling all this distributed and diverse information.

EMC’s pitch boils down to its view that Isilon’s native OneFS file system can do a better job handling big data than Hadoop’s native HDFS. “You can’t just say ‘big data’ and cross yourself. We’re fleshing out the story,” said Jeremy Burton, EVP with EMC, in an interview last week. “With Hadoop, we’ve got a distribution we’ve integrated with Greenplum to run queries across both [data sets.] Hadoop’s HDFS file system — it’s cheap and cheerful and useful. But at a petabyte scale, why not run OneFS? If you want to scale out to a petabyte, drop in Isilon and you’re done. If you need to add more nodes to the cluster, our file system automatically takes care of them.”

In this offering, Isilon’s OneFS integrates the Hadoop Distributed File System protocol so it can apply it to whatever massive amounts of information reside in its scale-out storage repository. EMC’s enterprise Hadoop distribution uses the MAPR file system.

The new Isilon-Greenplum product, available now, also shows that EMC has been busily integrating a series of acquisitions for an assault on the big data problem. EMC bought Isilon in July 2010 and Greenplum along with its analytic database  appliance, the following November.

You’re subscribed! If you like, you can update your settings

  1. Thanks for the information. I will be looking into this more.

  2. When dropping in another Onefs server does it also act as a compute node? That’s what Hadoop is about: bringing the computation to the data.

Comments have been disabled for this post