- A partition is a collection of tasks and associated resources grouped within a virtual processor according to the function of the tasks. There are multiple partitions within a single virtual processor. Partitions are the primary mechanism used by the database to manage parallel programs.
- Partitions are the subdivision of vproc software processes into 48 semi-isolated domains.
For example, in an AMP vproc, Partition 11 is the AMP Worker Task partition. In all other vproc types, Partition 11 is unused.
Another partition description is only meaningful in a dialog between client programs and the database. It has nothing to do with PDE vproc partitions, but is a way of enforcing rules about what a client session is allowed to do and of keeping client sessions isolated from each other. This concept of partitions is centered in the CLIv2 interface, specifically the CONNECT parcel.
- Partitions 0 through 6 are reserved by PDE
- Partitions 7 through 47 are for use by the database
The table listed under Partition Assignment Listing describes the individual partitions. The database uses the following vprocs:
|AMP||Access module processors perform database functions, such as executing database queries. Each AMP owns a portion of the overall database storage.|
|GTW||Gateway vprocs provide a socket interface to the database.|
|Node||The node vproc handles PDE and operating system functions not directly related to AMP and PE work. Node vprocs cannot be externally manipulated, and do not appear in the output of the Vproc Manager utility.|
|PE||Parsing engines perform session control, query parsing, security validation, query optimization, and query dispatch.|
|RSG||Relay Services Gateway provides a socket interface for communication with the Teradata Data Stream Utility.|
|TVS||Manages database storage. AMPs acquire their portions of database storage through the TVS vproc.|
For more information on partition usage, see Process Scheduling: CPU Utilization Columns.