This article is part of an Essential Guide, our editor-selected collection of our best articles, videos and other content on this topic. Explore more in this guide:
2. - On-site video interviews: Read more in this section
- Upgrades expected, but few takeoffs from the launch pad at EMC World
- George Crump on EMC's DSSD acquisition
- EMC ECS aims to take down Amazon, Google
- EMC World day one: DSSD holds attention
- DSSD acquisition leads SSD interest
- Matchett: HDFS offerings from EMC cater to several sweet spots
- McClure's take on Syncplicity updates
- George Crump on dedupe's effect on performance
- EMC's Emsley discusses modernization of data protection
- The 411 on ViPR's data services
Explore other sections in this guide:
Think of ViPR data services as plug-ins, analyst saysDate: May 16, 2014
All roads led back to ViPR 2.0 at EMC World 2014 -- or most of them, according to Greg Schulz, founder and senior analyst at IT consulting firm StorageIO. The trick for EMC customers, he said, is not to get bogged down in the software-defined storage marketing muck that's spread throughout much of the storage industry, and instead focus on the twofold benefits of ViPR 2.0. Storage pros should think about ViPR in terms of storage management, he explained.
"The piece that gets a little confusing to some people when it comes to ViPR are the data services," Schulz said. To understand the value of ViPR data services, think of them as plug-ins, or additional modules, such as the Hadoop Distributed File System (HDFS).
For example, he said, let's say you plug in HDFS and turn it on as a data service. "Now the control part of ViPR manages it -- the provisioning, the allocation, the tracking, the logging," Schulz explained. "But a separate piece, a separate node, actually runs the HDFS -- which means you can present an HDFS object, an HDFS data node, to the HDFS." The advantage, he said, is that instead of having local, direct-attached storage (DAS) in various Hadoop nodes, shared storage is accessed via the HDFS node.
"The MapReduce nodes access the HDFS server and shared storage from ViPR data services, similar to how they would from an Isilon [or other solutions] that presents HDFS shares using DAS," he said.
The management aspect of ViPR needs to be kept at the forefront of any conversation about the newest version, Schulz said. "The ViPR control plane, or management nodes, do the light work, only getting involved when needed for assisting underlying storage, then getting out of the way."
Schulz added: "Think of ViPR control nodes, which run as virtual machines [three for redundancy or five for very large sites] as your ideal manager. They are there when you need them to help without adding overhead, and then get out of the way, leaving you to get your work done other times. Or if you are into VMware, think of ViPR control plane nodes as similar to VMware vCenter. That adds value for management of vSphere ESXi hypervisor hosts, then gets out of the way, leaving the hypervisors to do the heavy lifting as needed."