| 摘要 |
Techniques for making aggregated entries in a database table which aggregate information from other entries in tables in the database system. The techniques permit the aggregated entries to contain not only metric values aggregated from the other entries by techniques such as averaging in which the individual values are lost, but also sets of individual values from the other entries. One area of application for the techniques is the roll up tables used in the management systems for database management systems to reduce the size of historic information about events that have occurred in the database management system. Each roll up entry in a roll up table is an aggregated entry that contains information about some number of events. A roll up entry that uses the techniques contains a representation of a set whose values are the occurrence times of the events that are represented by the rollup record. Among the techniques that can be used to represent the set of occurrence times are a comma list of the occurrence times and a bit map which has a bit for each second in a day. Roll up entries that contain such representations of sets of occurrence times may be analyzed to determine whether occurrences of events are related, and if they are, the fact of the relationship can be used to design filters that can be applied in the roll up process, in error reporting, and in the analysis of the roll up tables.
|
