EMC today sought to extend the reach and scope of its storage portfolio via the acquisition of a provider of shared, scalable Flash storage systems and extensions to its ViPR software-defined storage offering that adds support for both block data and commodity drives.
In addition, EMC formally unveiled the EMC Elastic Cloud Storage (ECS) Appliance that unifies file, object and HDFS support. Formerly known as Project Nile, the EMC Elastic Cloud Storage Appliance is designed to enable IT organizations to more easily build their own private clouds using petabytes of commodity storage.
Speaking at the EMC World 2014 conference, EMC chairman Joe Tucci says the acquisition of DSSD, a startup company still in stealth mode that EMC says is building “rack-scale” Flash storage systems using new storage interfaces, will drive applications based on in-memory databases such as SAP HANA along with next-generation cloud services. Due out in 2015, the Flash storage systems that EMC plans to bring to market are being developed by a team led by Andy Bechtolsheim, who was a co-founder of Sun Microsystems and most recently founder of Arista Networks.
Tucci says both DSSD and the EMC Elastic Storage Appliance are the latest examples of a federated EMC storage strategy that, unlike rival approaches, is not based on a single, one-size-fits-all architecture. All those different storage systems can be managed within the context of a single EMC ViPR controller, which is a core element of building a next-generation, software-defined enterprise. In fact, EMC says that going forward, ViPR 2.0 will be the core data plane from which all storage will hang. By adding support for the OpenStack Cinder plugin to make ViPR compatible with a broader array of third-party storage systems and commodity drives, EMC claims that ViPR 2.0 now addresses 80 percent of all existing storage capacity.
When it comes to building a software-defined enterprise, EMC contends that what will ultimately differentiate one storage architecture from another is the ability to manage data at scale inline. To back up that claim, EMC also announced a $1 million guarantee under which it is promising that EMC ExtremeIO Flash storage will not only not crash, but also will never throttle back on I/O in the face of contention for resources. EMC is also backing up that claim with a new trade-in program that will compensate IT organizations for swapping out existing Flash storage systems for EMC ExtremeIO systems.
Finally, EMC announced it is working with its VMware unit to make it easier to deploy hybrid clouds based on EMC storage and VMware virtual machines.
Despite the general emphasis on Flash, EMC notes that the vast majority will continue to run on magnetic storage for the foreseeable future. Not only is the cost per gigabyte of magnetic storage significantly lower than Flash storage, less than 10 percent of data is active enough to require Flash-level performance attributes.
Ultimately, EMC is making a case for a new federated storage architecture that will be required to support petabytes of data running across different classes of storage systems. The challenge will be to figure out how to manage all that data when no one is sure how much data is duplicated across the enterprise, nor are they sure where exactly that data should be located at any given time to drive optimal application performance.