Abstract

Web servers and large enterprises demand online backup capability to protect data that must be available continuously and eliminate the down-time needed to perform conventional off-line backups. Online backup demand is fueled by growing data capacities that have lengthened backup window time frames and by the significant loss in productivity that occurs when servers must be taken offline.

Snapshots allow applications to take online backup. This paper discusses the design and implementation of a layered device driver in Linux for persistent snapshots. This paper also discusses the design issues involved in developing a snapshot device driver in a clustered environment.