|
The data file is loaded in serial order, each record
getting a unique record number. A separate record is created in an "index" file
containing a key and the data record number (address).Features
- More than one index may be maintained (see bottom diagram)
- The index is kept in sorted order, most often Tree techniques (a B+ Tree being
the most common), are used for maintaining the order of a index. (diagram top).
Blocks are usually linked sequentially.
- The data file is loaded in serial order, each record getting a unique record
number. In order to improve index performance, the data file may be sorted, and indexes
regenerated, when not in use. Subsequent additions/deletions may destroy the sequence, so
periodic reorganisation is done to restore it.
- Access may be sequential or direct.
- Records have same length.
- This is the most common file format and is the basis for most Relational
databases.
|
