Recovering from Main-Memory Lapses

Recovery activities like logging checkpointing and restart are used to restore a database to a consistent state after a system crash has occurred Recovery related overhead is likely to form a bottleneck in a main memory database since I O activities are performed for the sole purpose of ensuring data durability In this paper we present recovery algorithms which reduce recovery related overheads in main memory databases The bene ts of our algorithms include the following Disk I O is reduced by logging to disk only redo records during normal execution The undo log is normally resident only in main memory and is garbage collected after transaction commit Checkpoints need not be transaction consistent uncommitted data can be written to disk by also writing out relevant parts of the undo log Contention on the system log is reduced by having per transaction logs in memory And nally the algorithms make only a single pass over the log during recovery Thus our recovery algorithms combine the bene ts of several techniques proposed in the past The ideas behind our algorithms can be used to advantage in disk resident databases as well