HP Endows Storage With Grid Computing Features

Hewlett-Packard is laying the foundation to move its storage to the next stage: grid computing.

Under a grid architecture, storage devices will act as intelligent nodes that can be turned off and on for instant scalability and easy management, said Hal Woods, CTO for storage area networks for HP’s StorageWorks unit, during a gathering of HP storage experts last week in Colorado Springs, Colo.

The approach could enable VARs to provide a single, integrated architecture that allows their customers to share storage devices and use them to quickly take advantage of new storage services by making those devices easier to install and reposition.

Dhruv Gulati, executive vice president at Lilien Systems, a Mill Valley, Calif.-based HP solution provider, said HP has already added elements of grid computing to recent products including its new enterprise NAS gateways, virtual tape libraries and Reference Information Storage System (RISS) archiving array. “It’s not a complex concept,” he said. “There are no barriers to the channel to sell it. And customers get cost-effective scaling and the ability to buy as they need and grow as they need.”

Woods said the storage grid will be made of intelligent nodes combining storage capacity, processing power and connectivity that can be plugged into the grid for an instant increase in performance. Those nodes, called Smart Cells by HP, can be clustered together with software that dictates services such as file serving, backup and so on, he said. In turn, those clusters can be managed as part of a single system, the grid, which also allows them to be repurposed or enhanced as needed, he said.

Woods contrasted this approach to current storage architectures based on storage arrays with fixed capacity, performance and connectivity. With the grid, customers could add iSCSI or Infiniband capabilities to a Fibre Channel array by just plugging the appropriate node into the grid, he said.

Grid computing should also allow users to repurpose storage hardware as needs change. “For example, at the end of the financial quarter, they could repurpose a virtual tape array as a disk array to increase performance temporarily and then [change] back to virtual tape later,” Woods said. “Or if a node in a cluster fails, they can repurpose another so they need fewer spares.”