alex_aldo - Fotolia

News Stay informed about the latest enterprise technology news and product updates.

EMC Isilon dives deeper into analytics, Hadoop, data lakes

EMC added a NAS array that can scale to 50 PB and upgraded its operating system to support the latest version of HDFS, plus OpenStack Swift.

EMC today upgraded its Isilon scale-out NAS platform to add analytics capabilities for big data, particularly Hadoop.

Isilon added its densest storage array, while upgrading its operating system to support the latest versions of Hadoop.

The EMC Isilon HD400 holds 3.2 PB per rack and scales to 50 PB in a cluster. The system is designed for customers requiring a minimum of 2 PB of storage. The HD-Series is new for Isilon, joining its S-Series (high performance), X-Series (high throughput) and NL-Series (archiving) platforms. The HD-Series is the highest-scaling Isilon system, eclipsing the NL-Series, which can store 31 PB in a cluster.

Isilon's upgraded OneFS 7.2 operating system supports Hadoop Distributed File System (HDFS) 2.3 and 2.4, as well as OpenStack Swift file and object storage.

Isilon added certification from enterprise Hadoop vendor Hortonworks, to go with previous certifications from Cloudera and Pivotal. Certification allows those vendors' analytics tools to run on Isilon.

"We call this being future-ready," Suresh Sathyamurthy, head of product marketing for EMC's Emerging Technologies division, said of the HD400. "If you know your data is doubling every two years and you have a scale-out architecture, you can buy a node and plug it in when your data grows."

EMC recommends Elastic Cloud Storage for object storage, but OpenStack Swift support for Isilon gives the NAS platform flexibility.

"This is our first step toward supporting object storage on Isilon," Sathyamurthy said. "You can input an object as a file and read it back as an object, and vice versa."

The increased Hadoop support is part of EMC's data lake strategy for unstructured data. The goal is to simultaneously support file, object and HDFS natively.

Hadoop was originally designed to run on DAS rather than shared storage. EMC is trying to persuade Hadoop users to replace DAS with Isilon. With native HDFS support, Isilon can act as the storage layer in the Hadoop cluster.

Isilon has supported HDFS natively since 2012.

"Today with Hadoop you have to move data from primary storage to a DAS cluster and create three copies of data," Sathyamurthy said. "With HDFS natively integrated, you can do in-place analytics without having to move data and you don't need a DAS cluster. Hadoop purists believe Hadoop should be on a DAS cluster, but that's because nobody else has this functionality."

Ashish Nadkarni, an IDC storage research director, said EMC is positioning Isilon to go after "rogue open source deployments" of Hadoop that do not use networked storage.

"EMC's data lake strategy is about making Isilon friendly for Hadoop deployments so you don't need a separate Hadoop cluster," Nadkarni said. "You can make Isilon your Hadoop storage."

Next Steps

Can Hadoop be used without HDFS?

EMC Isilon gets deduplication

Dig Deeper on Storage management and analytics