The process of undoing uncommitted changes that an application process has made. A backout may be necessary if an application process fails or in a deadlock situation.
See also [roll back]
backup
A copy of a database or table space that can be stored on a different medium and used to restore the database or table space in the event of failure or damage to the original.
backup pending
The state of a database or table space that prevents an operation from being performed until the database or table space is backed up.
base table
A table created with the CREATE TABLE statement. A base table has both its description and data stored in the database.
before trigger
A trigger that is activated before the triggering SQL operation has completed. The triggering operation can be a fullselect, SIGNAL SQLSTATE, or SET transition variable SQL statement.
BID
See [block identifier]
binary large object (BLOB)
A sequence of bytes with a size from 0 bytes to 2GB minus 1 byte. This string does not have an associated code page and character set. BLOBs can contain image, audio, and video data.
bind
To convert the output from the SQL compiler to a usable control structure, such as an access plan, application plan, or package. During the bind process, access paths to the data are selected and some authorization checking is performed.
bind file
A file the precompiler produces when the PRECOMPILE command or the respective API is used with the BINDFILE option.
bit data
Data with character type CHAR or VARCHAR that is not associated with a coded character set and therefore is never converted.
BLOB
See [binary large object]
block
(1) A string of data elements that is recorded or transmitted as a unit.
(2) A set of contiguous data pages in a buffer pool.
(3) A set of consecutive pages on disk.
block-based I/O
A database manager method of reading contiguous data pages from disk into contiguous portions of memory.
block fetch
A DB2 function that retrieves (fetches) a large set of rows together. Using block fetch can significantly reduce the number of messages sent across the network. Block fetch applies only to cursors that do not update data.
block identifier (BID)
An entry that is stored along with a key value in the leaf node of a block index. This identifier references a particular block in a multidimensional clustering table.
block index
An index that is structured in the same manner as a traditional record identifier (RID) index, except that at the leaf level the keys point to a block identifier (BID) instead of a RID.
block locks
The locking of a block within a multidimensional clustering environment.
block map
A bitmap that contains an array of block states, one for each block in the multidimensional clustering table.
block size
Specifies the number of pages in a block and is equal to the extent size. Also known as block factor.
blocking
An option specified when binding an application. It allows caching of multiple rows of information by the communications subsystem so that each FETCH statement does not require the transmission of one row for each request across the network.
buffer manipulators
The processes used in backup and restore operations to read from or write to the database. By default, a single buffer manipulator is used; however, this can be overridden by using the parallelism option of the BACKUP DATABASE or RESTORE DATABASE commands.
buffer pool
An area of memory into which database pages are read, modified, and held during processing.
built-in function
A strongly typed, high-performance function that is integral to DB2 UDB. A built-in function can be referenced in SQL statements anywhere that an expression is valid.
