Petabyte Storage Infrastructure Project

Scaling to New Heights

The Petabyte Storage Infrastructure Project will provide

1. low-cost, petabyte-scale, generic storage via the use of replicated commodity components, tape-less (i.e., disk-to-disk) backup, and a high level of administrative automation
2. a fast cache, to support large computations with intensive local storage
3. thousands of environmental sensors to support experimenting with collecting and storing large sensor-derived environmental data sets.

Motivation and Goals

• Where is data coming from?
• The Structure of Generic Storage Costs
• Existing Petabyte Stores
• Building a Lower Cost Generic Object Store
• An eScience Data Cache
• A Sensor Infrastructure
• Applications

Current Designs and Implementation

• Powerpoint Overview (3.5MB)
• Archival Store
• Wireless Mote Testbed

Technologies

• Oceanstore

Related Efforts

• CITRIS
• Millennium
• TinyOS
• Self-* Storage Systems

About the Project

• People