This job type runs a loop in which all subjobs are executed in parallel. It can be used for groups of independent jobs and load to separate cubes / cube slices.
Fail on status
If the job executes several loads or sub-jobs, the selected option defines the behavior in case of a warning or an error message in one of the loads or sub-jobs. The options are described below.
| none | All subsequent loads or sub-jobs are executed even if errors or warnings occur. The job terminates with "Completed with warnings" or "Completed with errors" or "Completed successfully". |
| error | In case of an error message, the job terminates without executing subsequent loads or sub-jobs and the job terminates with status "Failed". In case of warnings subsequent loads or sub-jobs are executed and the job terminates with "Completed with warnings". |
| warning | In case of a warning or an error message, the job terminates without executing subsequent loads or sub-jobs and the job terminates with status "Failed". |
| inherit | If the job is executed directly (without parent job) it uses failOnStatus "error". Otherwise if the job is used as a sub-job it inherits the failOnStatus of its parent job (see corresponding descriptions of these failOnStatus options above). |
Example
variables: level1
Loop source: level1 receiving Jan | Feb | Mar
Execution: Load OneMonth
In the LoopParallel job, the load OneMonth will be executed 3 times simultaneously:
- with variable level1="Jan"
- with variable level1="Feb"
- with variable level1="Mar"
Limits of parallelism are globally configured:
| Parameter | Description of parameter value | Default value | Type |
|---|---|---|---|
| maxParallelWritelExecutions | Maximum number of parallel root executions without counting the dependent subjobs / loads of a parallel running execution. Write executions are those that may change data in the target system, such as starting a load or a job. | 5 | integer |
| maxParallelReadExecutions1 | Maximum number of parallel running Integrator executions. Read executions normally only read data from source systems and don't change data in target systems. Examples: data preview, function preview, tests, drillthrough, metadata, column output. | 7 | integer |
| maxParallelSubExecution | Maximum number of parallel sub-executions inside of a parallel execution | 5 | integer |
Notes
-
In the case that there is already another running parallel execution at the beginning of a new parallel execution, the new one will only have one parallel sub-execution, i.e., no effective parallelism happens inside the execution. Therefore, as the best practice, at any given moment, run only one execution of a LoopParallel job or a Parallel job with multiple parallel sub-executions. For this "interior parallelization" the limit of maxParallelSubExecutions (default: 5) applies.
-
Other parallel jobs that do not have parallel sub-executions can be efficiently executed simultaneously. For this "exterior parallelization" the limit maxParallelWriteExecutions (default: 5) applies.
Updated April 8, 2025