Range Buckets Registry entry

System Administrators have access to tools to tune the range indexing used by EMu. Support for automatic optimisation of range indexes is also provided. With these tools it is possible to achieve optimal range indexes with significantly faster range based searches in EMu.

Range indexing is really a series of "mini" indexes on a per field basis. Unlike the Two Level scheme, where all search terms are placed in the one index, range terms are placed in per field indexes that are then concatenated to form one range index. Each field index consists of a number of range buckets. These buckets are used to indicate whether a given value falls within the bucket or not.

There are two related considerations when establishing range buckets:

1. Data distribution: as best as possible data should be distributed evenly between the buckets.
2. Logical query ranges. The aim is to minimise the necessity to check a bucket for a value as checking whether records in a bucket match the query takes time; therefore if users are likely to search on particular ranges (decades for instance: 1/1/1910 to 1/1/1920), it makes sense to configure range buckets appropriately.

The setting of range bucket values is performed on a per field basis. Each range indexed field may have a Registry entry that defines the range buckets for that field. If a Registry entry does not exist, the built-in values defined when the table was designed are used.

Note: See Range Indexing for details about EMu's range indexing tools, methods and theory.