The job system supports two different operational models with respect to job runs, each one providing unique flexibility and applicability to different situations.
Multiple Execution Runs: You submit a job that has a target list of N targets where N may be any number. When the job is submitted, it creates N independent executions per run, each run operating on one target.
Single Execution Runs: You submit a job with a target list of N targets, but it only creates one execution per run.
As shown in the following graphic, a job runs against three targets: T1, T2 and T3. The job is run daily. When the job is submitted, it creates three executions: E11 on target T1, E12 on target T2, and E13 on target T3. The set of the executions (E11, E12, and E13) constitute Monday’s job run.
On Tuesday the job runs again and three more executions are created (E21,
E22, and E23), with each execution operating independently on a different target
(T1, T2, and T3). This set of the executions would be Tuesday’s job run.
As with the previous example, the job is run against three targets. However, with single execution runs, each run consists of only one execution: On Monday you just one have execution (E1) that operates on all three targets. On Tuesday you have another execution (E2) that again operates on all three targets.
Use a multiple-execution run when there are no dependencies between targets/tasks. Typically, when you create jobs that can operate on multiple targets in a completely independent manner, you should define them as multiple execution runs. For example,if you create an OS command job, when you submit the job against multiple targets using a command parameter such as “ps -ef | grep oracle” , the command is executed independently on all the targets: Successful completion of every execution is not required for the job to complete the run. This job model would be useful in a situation where, for example, a Management Agent for one of the targets is down and the failure of a single execution will not prevent the job from running.
Use a single-execution run when there are dependencies between targets/tasks.
You define a job using a single-execution run model when you want the job to
perform dependent tasks. For example, you define a job where you want it to
perform task1 on the first target, and depending on the success or failure of
that task, you perform task2 on the second target, and depending on the success
or failure of task2, the job performs task3 on the third target. In situations
such as this, where you have a job workflow consisting of dependent tasks, you
want to create a job based on the single-execution run model.
Single vs. Multiple Execution Runs
Enterprise Manager Information Roadmap
Copyright © 1996, 2007, Oracle. All rights reserved.