| 摘要 |
Database performance assessment typically entails logging and capturing utilities which tend to gather large quantities of unwieldy data leading to a cumbersome and indefinite analysis. A system and method for database performance analysis includes periodic sampling of pending database requests, rather than exhaustive monitoring and capturing all database access traffic, to identify areas of contention. A database access queue is periodically sampled, or scanned, to gather a snapshot of pending requests of database transactions. Pending requests are aggregated by an aggregating process which aggregates, the samples with previous samples corresponding to the same transaction. Correlating the aggregated samples identifies transactions which have been pending the longest and identifies tables and segments in the database which have a relatively high number of pending transactions. By periodically sampling, rather than exhaustively logging all requests, CPU intrusiveness is minimized, trivial, benign transactions are eliminated from the output, and the most burdened tables and segments are identified to enable a database administrator to make informed decisions about remedial actions. |
