News

posted on 14 March 2008 16:00

A paper published by a NetApp technical director and University of California researchers suggests that a MAID array with metadata stored in non-valatile RAM (NVRAM) could replace tape archives.

The authors believe that Pergamum could "realize significant cost savings by keeping the vast majority, as many as 95 percent, of the disks spun down while still providing reasonable performance and excellent reliability. "

The paper, entitled 'Pergamum: Replacing Tape with Energy Efficient, Reliable, Disk-Based Archival Storage' and available here identifies the need for low-cost archival storage with adequate performance for retrieval and auditing purposes. Costly tape libraries with offline tapes don't fulfill that need.

Nor do MAID (massive array of idle disks) arrays from suppliers like Copan and Nexsan because most of the disks are spun down and high-performance controller/servers are needed. Arrays of constantly spinning disks have too high energy requirements as well as also needing expensive controllers.

The authors' proposal is: "a distributed network of intelligent, disk-based, storage appliances that stores data reliably and energy-efficiently. While existing MAID systems keep disks idle to save energy, Pergamum adds NVRAM at each node to store data signatures, metadata, and other small items, allowing deferred writes, metadata requests and inter-disk data verification to be performed while the disk is powered off."

In a little more detail: "Pergamum takes an approach similar to that used in high-performance scalable storage systems [48, 36, 46], and is built from thousands of intelligent storage appliances connected by high-speed networks that cooperatively provide reliable, efficient, long-term storage. Each appliance, called a Pergamum tome, is composed of four hardware components: a commodity hard drive for persistent, large-capacity storage; on-board flash memory for persistent, low-latency, metadata storage; a low-power CPU; and a network port."

"Each appliance runs its own copy of the Pergamum software, allowing it to manage its own consistency checking, disk scrubbing and redundancy group responsibilities. Additionally, the CPU and extensible software layer enables disk-level processing, such as compression and virus checking. Finally, the use of standardized networking interfaces and protocols greatly reduces the problem of maintaining complex chains of dependent hardware."

It's interesting that Kaladhar Voruganti, a technical director in Net App's Advanced Products Group (ATG) is one of the authors of the paper. Just possibly it may have a bearing on NetApp's thinking about how best to archive data.

[Chris Mellor.]