Like many information technology tasks, database-related tasks can often be can be broken up and executed in parallel in order to speed up the task. This process of subdividing tasks for execution by multiple processors is called parallelism.

A task can be performed in parallel in several ways. The nature of the task, the database configuration, and the hardware environment determine if and how a system will perform a task in parallel, as well as the potential reduction in time required. Both I/O parallelism and query parallelism are possible.

I/O Parallelism

Parallel I/O means reading from or writing to two or more I/O devices at the same time in order to minimize required time. Parallel I/O can provide a significant improvement to I/O throughput.

Query Parallelism

The two basic types of query parallelism are inter-query parallelism and intra-query parallelism.

Inter-query parallelism allows multiple applications to query a database simultaneously. The queries will execute independently, but the system will execute all of the queries at the same time.

Intra-query parallelism is the processing of several parts of a single query simultaneously using either intra-partition parallelism or inter-partition parallelism.

Not all queries are suitable for parallel processing. Fast-running queries do not need parallel processing, for example. That’s why it’s essential for database administrators to carefully consider what types of devices and queries need parallel processing to improve performance.

SQL Server software leader Idera understands the importance of parallelism, and Idera Senior Product Consultant Robert Davis recently recorded a webcast titled “Performance Implications of Parallelism” that provides some useful perspective on the subject.