|
Partitioning lets you achieve a few things:
1. Higher availability
2. Better performance
When you partition a table, you are saying that part of the rows of the
table will go into one partition, another set of rows in another
partition, and so on.
Higher availability is achieved when you place these different
partitions on different disk devices. Should you lose a disk device,
then only attempts to access that partition will fail. If your
application accesses other partitions that are still available, the
application will not notice. Regular tables do not enjoy this feature.
Better performance is achieved by one of two ways. First, you can
spread the table's data among multiple disk devices by placing each
partition in a different tablespace on different disk devices. More
disk devices means more disk spindles at work to satisfy your I/O
requests.
Better performance is really noticable with a concept of "partition
pruning." If your table has three partitions, PART_A, PART_B, and
PART_C and each partition contains 1 million rows of data, then when
you query looks for data that is in only PART_B, the system
automatically knows not to look in the other partitions. Instead of
reading three million rows, you only have to read one million.
I would recommed reading the Oracle Concepts Guide in your
documentation for more information as partitioning can be a large
topic.
|
