DB2 10.5 for Linux, UNIX, and Windows
DB2 10.5 for Linux, UNIX, and Windows
The decision to create a multi-partition database must be made before you create your database. As part of the database design decisions you make, you will have to determine if you should take advantage of the performance improvements database partitioning can offer.
In a partitioned database environment, you still use the CREATE DATABASE command or the sqlecrea() function to create a database. Whichever method is used, the request can be made through any of the partitions listed in the db2nodes.cfg file. The db2nodes.cfg file is the database partition server configuration file.
Except on the Windows operating system environment, any editor can be used to view and update the contents of the database partition server configuration file (db2nodes.cfg). On the Windows operating system environment, use db2ncrt and db2nchg commands to create and change the database partition server configuration file
Before creating a multi-partition database, you must select which database partition will be the catalog partition for the database. You can then create the database directly from that database partition, or from a remote client that is attached to that database partition. The database partition to which you attach and execute the CREATE DATABASE command becomes the catalog partition for that particular database.
The catalog partition is the database partition on which all system catalog tables are stored. All access to system tables must go through this database partition. All federated database objects (for example, wrappers, servers, and nicknames) are stored in the system catalog tables at this database partition.
If possible, you should create each database in a separate instance. If this is not possible (that is, you must create more than one database per instance), you should spread the catalog partitions among the available database partitions. Doing this reduces contention for catalog information at a single database partition.
When you create a database, it is automatically created across all the database partitions defined in the db2nodes.cfg file.
When the first database in the system is created, a system database directory is formed. It is appended with information about any other databases that you create. When working on UNIX, the system database directory is sqldbdir and is located in the sqllib directory under your home directory, or under the directory where DB2® database was installed. When working on UNIX, this directory must reside on a shared file system, (for example, NFS on UNIX platforms) because there is only one system database directory for all the database partitions that make up the partitioned database environment. When working on Windows, the system database directory is located in the instance directory.
Also resident in the sqldbdir directory is the system intention file. It is called sqldbins, and ensures that the database partitions remain synchronized. The file must also reside on a shared file system since there is only one directory across all database partitions. The file is shared by all the database partitions making up the database.
Configuration parameters have to be modified to take advantage of database partitioning. Use the GET DATABASE CONFIGURATION and the GET DATABASE MANAGER CONFIGURATION commands to find out the values of individual entries in a specific database, or in the database manager configuration file. To modify individual entries in a specific database, or in the database manager configuration file, use the UPDATE DATABASE CONFIGURATION and the UPDATE DATABASE MANAGER CONFIGURATION commands respectively.
The database manager configuration parameters affecting a partitioned database environment include conn_elapse, fcm_num_buffers, fcm_num_channels, max_connretries, max_coordagents, max_time_diff, num_poolagents, and start_stop_time.