Capacity planning
You plan the use of disk, volume group, and file system resources to create an optimal operating environment for IBM® InfoSphere® Information Server.
As part of capacity planning for your installation, size your file systems and databases to accommodate your anticipated needs. After the installation, continue monitoring the file systems to ensure that sufficient space is available.
Before you change your computers, back up those computers where you plan to install InfoSphere Information Server.
File systems
Plan for the following separate file systems:
- Temporary storage
- Directories for scratch data that is used during processing and later discarded.
- Program code
- Program code directories for the InfoSphere Information Server installation directory, database server installation directory, and the IBM WebSphere® Application Server installation directory.
- Storage directories
- Database storage directories for the metadata repository database, and for the IBM InfoSphere Information Analyzer databases and IBM InfoSphere QualityStage® Match Designer database (if you are installing these product modules).
Isolate each functional subgroup or components in its own separate file system to accommodate different performance and capacity characteristics and backup requirements. File systems need to be expandable without requiring destruction or recreation, if the operating system that you are installing on permits it.
RAID and SAN configurations
Performance of a job (for IBM InfoSphere DataStage®, IBM InfoSphere QualityStage, and IBM InfoSphere Information Analyzer) depends on all components being optimized. For RAID (Redundant Array of Independent Disks) and SAN (system area network) configurations, maximum performance is a combination of maximum bandwidth (controllers and disk) and minimized contention.
The following guidelines can assist in the configuration of RAID or SAN technologies:
- Ensure that your database is optimally tuned.
- Ensure that computers where you install the repository tier and services tier are on a high-speed network.
- Minimize contention between temporary systems (scratch, buffer, and sort) and data file systems. For temporary storage, do not use a network file system (NFS). Consider using the local disk for temporary storage.
- Minimize contention between the disks and controllers that are associated with InfoSphere Information Server file systems and other applications or servers.
- Consider isolating multiple mount points to separate high-speed disk interconnects and controllers.
- Consider the trade-off between the granularity of file systems and the underlying configuration in comparison to the available, unused storage.
- Do not create overly complex device configurations. These configurations can be difficult to administer and might not offer noticeable performance improvement.
- If possible, check your configuration with experts in hardware, storage, operating systems, and InfoSphere Information Server.
- Systems with parallelism require different expertise than systems with engines that do not feature parallelism.
The optimal disk configuration balances cost, complexity, ease of administration, and ultimate performance.