DB2 Partition Groups

A database partition group is a set of one or more database partitions. By grouping database partitions, you can perform database operations at the partition group level rather than individually on each partition. This allows for database administration flexibility. For example, let's say you want to create a buffer pool with the same definition in three partitions.

If you first create a partition group pgall that consists of the three partitions, you can associate the buffer pool bp1 you are about to create with this partition group. This lets you use the same buffer pool definition on each partition.
Partition groups also allow you to associate table spaces to database partitions. For example, if you would like table space tbls1 to use only database partitions 1 and 2, you can create a partition group pg12 with these two partitions, and then associate the table space tbls1 to pg12.


Buffer pool definition is repeated across all the partitions. If you create the buffer pool with 20,000 4K pages, each partition will allocate 20,000 4K pages. Note that the 20,000 4K pages are not split across all three partitions as the figure may suggest.
To create a partition group, use the CREATE DATABASE PARTITION GROUP statement.

When you create a database, three partition groups are created by default.

• IBMCATGROUP is the partition group where the DB2 catalog table space (SYSCATSPACE) resides. It consists of only one partition, the one where the CREATE DATABASE command is issued.
• IBMTEMPGROUP is the partition group where the system temporary table space (TEMPSPACE1) resides. It spans all partitions.
• IBMDEFAULTGROUP is the partition group where the user table space (USERSPACE1) resides. It spans all partitions.

Created with Artisteer