Check out the new USENIX Web site. next up previous
Next: Persistence. Up: Recovery Previous: Data stored on disk.

Warming the cache.

Hummingbird lacks a directory structure and all meta-data consistency dependencies are contained within clusters. There is no need for an analog of the UNIX fsck utility to ensure file system consistency after a crash. During a planned shutdown, Hummingbird will write the file-to-cluster mappings to disk. (They are also written periodically.) During a crash, the system has no such luxury. So, while the file system is guaranteed to be consistent with itself, the lack of directories might make it impossible to locate files on disk immediately after a crash if the file-to-cluster mapping was not up to date. Hummingbird speeds up crash recovery time with a log containing the cluster identifiers of popular clusters. The sync_hot_clusters_daemon() creates this log using the cluster LRU list and writes it to disk periodically. After a crash, these clusters are scanned in first, quickly achieving a hit rate close to that before the crash.

Liddy Shriver 2001-05-01