RelationalDBDesign RelationalDBDesign


Parallel Server   «Prev  Next»
Lesson 7 Node and disk affinity
ObjectiveDefine node and disk affinity.

Node | Disk affinity (OPS system)

In the previous lesson, we discussed how to set up and use instance affinity. In this lesson, we take a look at a different concept: disk affinity. The OPS system contains a group of interconnected instances and supporting data files stored on disks. If an instance is located on the same node as a data file, the two are considered to have node affinity[1]. The idea is expanded further to disk affinity[2], which means that a database instance is located on the same node as a disk that contains the data file.
The Optimizer uses disk affinity to decide which instance runs each component of a parallel query. The Optimizer looks at the data and finds the instance closest to the partition that it needs for the portion of the parallel transaction that the instance is to run. This improves response time by causing the data access to be done as much as possible within a local node, reducing traffic on the network.
Note: Disk affinity is important for parallel DML, parallel queries, and all parallel operations in OPS systems.
The next lesson covers how to set up Transparent application failover.
[1] Node affinity: A database instance of an Oracle Parallel System has node affinity with a data file when the data file and the database instance are located on the same computer (also called the node).
[2]Disk affinity: A database instance of an Oracle Parallel System has disk affinity with a data file when the data file is located on a disk that is mounted on the same computer as the database instance.