Check out the new USENIX Web site. next up previous
Next: Synchronous Upgrade and Rollback Up: Asynchronous Upgrade and Rollback Previous: Packages

File System Checkpoints

An alternative approach, that addresses the speed concern (while maintaining the other goals), is to use file system checkpoints. To upgrade the software, we shut down service on a node, checkpoint the filesystem, and then restart service. To roll back the software, we shut down service, roll back the filesystem, and then restart service. Because this approach avoids the need to execute uninstallation and installation code, it may better meet the speed goal.



mukesh agrawal 2003-06-17