Storing Stuff Forever

Mary Baker, Hewlett-Packard, Palo Alto

Stevenson Hall 1300
11:00 AM - 11:50 AM

Many enterprises, organizations and individuals find themselves needing to preserve large volumes of quickly accessible digital content indefinitely into the future. The costs of doing so are often prohibitive, and even when money isn't a problem, lots of traditional storage systems and processes aren't designed with good ideas about how to safeguard these digital assets over long time periods. We examine threats to long-lived data from an end-to-end perspective, taking into account not just hardware and software faults but also faults due to people and organizations. We present a simple model of long-term storage failures that helps us reason about various strategies for addressing some of these threats. Using this model we are building tools that exploit the most important strategies for increasing the reliability of long-term storage: detecting latent faults quickly, automating fault repair to make it cheaper and faster, and increasing the independence of data replicas.