Partitioned Tables
In a VLDB environment, partitioned tables help to make the database more available and
maintainable. A partitioned table is split up into more manageable pieces, called partitions, and
can be further subdivided into subpartitions. The corresponding indexes on partitioned tables can
be nonpartitioned, partitioned the same way as the table, or partitioned differently from the table.
Partitioned tables can also improve the performance of the database: Each partition of a
partitioned table can be accessed using parallel execution. Multiple parallel execution servers can
be assigned to different partitions of the table or to different index partitions.
For performance reasons, each partition of a table can and should reside in its own tablespace.
Other attributes of a partition, such as storage characteristics, can differ; however, the column
datatypes and constraints for each partition must be identical. In other words, attributes such as
datatype and check constraints are at the table level, not the partition level. Other advantages of
storing partitions of a partitioned table in separate tablespaces include the following:
It reduces the possibility of data corruption in more than one partition if one tablespace is
damaged.
Each partition can be backed up and recovered independently.
You have more control of partition??“to??“physical device mapping to balance the I/O load.
Even in an ASM environment, you could place each partition in a different disk group; in
general, however, Oracle recommends two disk groups, one for user data and the other
for flashback and recovery data.
Pages:
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834