DB2 for z/OS: Data Sharing in a Nutshell

Chapter 2. Data sharing architecture 37

Most of these standards apply equally to non-data sharing environments.

Commit frequency has additional implications in a data sharing environment.

Long running update, insert, and delete processes lock resources that other

processes may need to access. Also, inadequate commit frequency for long

duration application processes — for example, batch jobs — may have the

greatest impact on data sharing performance by reducing lock avoidance. Refer

to 7.6, “The problem of runaway threads” on page 128.

Lock avoidance is the ability of DB2 to return rows to the application without

holding a lock on the data row or page. Avoiding locks locally is good for

performance; avoiding global locks can have a significant ...