Scalable Database Server, HiRDB Version 8 Description
Table data on disk normally is read from a global buffer (or a local buffer) one page at a time. However, this data can also be read in batches of multiple pages, rather than one page at a time. This capability is provided by the prefetch facility. By using the prefetch facility, you can reduce the number of I/O operations between the table data on disk and the global buffer (or local buffer).
- Organization of this subsection
- (1) Application criteria for the prefetch facility
- (2) Specifying the prefetch facility
(1) Application criteria for the prefetch facility
The prefetch facility is available when the following conditions are satisfied:
- The accessed data is in a HiRDB file system area for which the raw I/O facility is used.
- A large data set is being manipulated.
- One of the following SQL statements is being executed:
- A SELECT, UPDATE, or DELETE statement that does not use an index
- A SELECT, UPDATE, or DELETE statement that performs an ascending search* using an index or cluster key (except for = and IN conditions)
- * For multi-column indexes, searching is performed in the order specified by the index definitions.
(2) Specifying the prefetch facility
- For a global buffer
- The prefetch facility is enabled by specifying a value of at least 1 for the -m option of the pdbuffer operand. The specification in the -p option of the pdbuffer operand is of the number of pages to be fetched at a time. For details about the pdbuffer operand, see the HiRDB Version 8 System Definition.
- For a local buffer
- The prefetch facility is enabled by specifying the number of pages to be fetched at a time in the -p option of the pdlbuffer operand. For details about the pdbuffer operand, see the HiRDB Version 8 System Definition.
All Rights Reserved. Copyright (C) 2007, Hitachi, Ltd.