Oracle Parallel Processing

Chapter 3. Parallel Execution Concepts

This chapter presents Oracle’s parallel execution (sometimes known as parallel SQL) features: parallel query, parallel DML, parallel DDL, and several other types of parallel operations. If implemented effectively, parallel operations can significantly improve the performance of data-intensive jobs. The decision to execute an operation in parallel is made by Oracle at runtime and is mostly transparent to programmers and end users.

Oracle’s parallel execution features are different from the features provided by Oracle Parallel Server (OPS). You don’t need OPS to perform parallel SQL execution. However, parallel SQL execution can work in conjunction with OPS, and some aspects of parallel execution apply only to Oracle Parallel Server. Throughout this chapter, unless we note otherwise, we are dealing with parallel SQL operations in a standalone instance environment. Part III of this book discusses OPS; in particular, Chapter 13, deals with parallel execution in an OPS environment.

What Is Parallel SQL?

When a user connects to an Oracle instance, a process referred to as a server process, or shadow process, starts on the database server. The purpose of this shadow process is to perform the operations requested by the user process. When the user executes a SQL statement, the corresponding shadow process performs the necessary tasks to get the result. In a serial execution environment, all the tasks are performed by a single process (the shadow process), ...