EMC VPLEX Pushes Storage Virtualization Toward Private Clouds


EMC on Monday unveiled its VPLEX technology for virtualizing storage over long distances, promising the ability to eventually migrate data over any expanse and help customers adopt private cloud-based infrastructure.

VPLEX enables customers to share storage over great distances and build a truly dynamic data center, said Pat Gelsinger, president and COO for EMC's infrastructure products.

The Hopkinton, Mass.-based vendor introduced VPLEX just prior to the official opening of its EMC World technical and end-user conference, which is being held this week in Boston.

Brian Gallagher, president of EMC’s Storage Virtualization Product group, said that storage requires two components. The first is tiering, such as what EMC offers using its FAST (fully automated storage tiering) technology. The second is federation, which is the ability to easily manage all the storage within a data center.

With VPLEX, EMC is taking those two concepts to global data centers, Gallagher said.

VPLEX is good news for customers trying to define cloud computing, said Keith Norbie, vice president of sales at Nexus Information Systems, a Minnetonka, Minn.-based solution provider and EMC partner.

“People are a long way from the cloud now,” Norbie said. “But they’re eager for it.”

The majority of the world has fragmented data centers, Norbie said.

“They want to connect and extend those data centers,” he said. “VPLEX lets them embrace heterogeneous storage and the cloud. And most customers have heterogeneous storage, either for tactical, strategic, or legacy reasons.”

VPLEX is a series of appliances that allow access to EMC and non-EMC storage with a single product. It is array-aware, which Gallagher said lets customers leverage the investment they have already made in their storage infrastructures by taking advantage of existing storage array features.

Each VPLEX node, or engine, has two processor boards, and features the ability to tie the storage from multiple EMC and non-EMC storage arrays in such a way that data can be dynamically migrated from one to another, Gallagher said.

With array-aware abilities, VPLEX allows caching, replication, and automation between multiple arrays, and will over time support other vendors’ APIs, Gelsinger said. EMC also plans to open its APIs to other vendors as well, he said.

Next: Expanding VPLEX To Build Private Clouds

VPLEX will come in four versions: VPLEX Local, VPLEX Metro, VPLEX Geo and VPLEX Global, Gallagher said.

VPLEX Local allows migration of data between multiple arrays in a local data center in a non-disruptive fashion, making it suitable for such tasks as technology refreshes. VPLEX Local scales to up to four VPLEX engines.

VPLEX Metro allows data to be accessed across two data centers up to about 100 km apart. With VPLEX Metro, customers can provision and share data across the two data centers and take advantage of excess capacity in one data center. It scales to up to eight VPLEX engines, four at each site.

“When combined with VMware’s VMotion, you can non-disruptively migrate virtual machines, applications, and data associated with those applications,” Gallagher said.

For example, in a disaster recovery scenario, if a customer experiences a disaster in one data center it can migrate the applications and data to the other, he said.

Both VPLEX Local and VPLEX Metro are available starting this week.

Next year, EMC plans to expand its VPLEX infrastructure with two new solutions, Gallagher said.

The first, VPLEX Geo, will allow asynchronous access to data between two data centers at any distance.

The second, VPLEX Global, will allow multi-site configurations, and link multiple data centers into a single large virtual data center, he said.

VPLEX is different from EMC’s InVista storage virtualization appliance, which has been around for several years, Gallagher said. InVista, which will remain in production, virtualized the storage from multiple storage arrays into a single pool, while VPLEX allows the migration of data between multiple arrays.