Energy Efficient, Reliable, Disk-Based Archival Storage

Pergamum: Energy Efficient, Reliable, Disk-Based Archival Storage
Ethan Miller [Professor of Computer Science, UC Santa Cruz]
Abstract:

As the world moves to digital storage for archival purposes, there is an increasing demand for reliable, low-power, cost-effective, easy-to-maintain storage that can still provide adequate performance for information retrieval and auditing purposes. Unfortunately, no current digital archival system - tape, disk, or optical disk - adequately fulfills all of these requirements. To address this challenge, we developed Pergamum, which stores data in a network of "bricks", each of which contains a disk, low-power CPU, and flash memory. In normal operation, only a small fraction of the disks in a Pergamum system are spun up, dramatically reducing power consumption. Since the other components use very little power, a Pergamum system with 5% of the disks in use consumes less than 2 watts per terabyte of stored data.

Pergamum is also designed to be very highly reliable. Pergamum stores additional redundant information on each brick, allowing the brick to repair itself in many cases, and takes advantage of the bricks' network connections and low-powered flash memory to constantly verify storage integrity between bricks without the need to constantly spin up disks, ensuring data preservation with low power requirements. If an error is found, Pergamum can rebuild the lost data with low peak power consumption, avoiding the need for large-scale power and cooling infrastructure. Pergamum is also designed to evolve over time: it is based around standard IP-based networking protocols, and can accommodate bricks using any storage technology, as long as the bricks "speak" the standard protocols. By using these techniques, Pergamum can provide archival storage at cost and power consumption comparable to tape with far better performance, reliability, and evolvability than a tape-based system. Current research on Pergamum is exploring issues with integrating new devices and retiring old devices, failure detection in a system with hundreds of thousands of devices, and search across exabyte-scale storage in a power-constrained environment.




Bio:

Ethan L. Miller is a Professor in the Computer Science Department at the University of California, Santa Cruz, where he is the Associate Director of the Storage Systems Research Center (SSRC). He received an Sc.B. from Brown University in 1987 and a Ph.D. from UC Berkeley in 1995, where he was a member of the RAID project. His current research projects, which are funded by the National Science Foundation, Department of Energy, and industry support for the SSRC, include long-term archival storage systems, scalable metadata systems, file systems for non-volatile memory technologies, reliable and secure storage systems, and issues in petabyte-scale storage systems. Prof. Miller's broader interests include file systems, operating systems, parallel and distributed systems, information retrieval, and computer security. Additional information is available at http://www.cs.ucsc.edu/~elm/ and http://www.ssrc.ucsc.edu/.

Category:

Science & Technology

Tags:

• CITRIS
• UCB
• Science
• Technology
• Research

[DrDargh]

10 years is up lets knock down this strip mall and rebuild it, such a miami attitude