In a new configuration, default options for data groups result in all data groups using the default shared object send (OBJSND) process for the system. Sharing an object send job typically reduces CPU usage on the source system.
However, if performance is slow or the object send process has a persistent backlog, it may be necessary to reduce the number of data groups sharing the object send process so that the process can keep up with the volume from its shared data groups. You can change a data group configuration to use a different shared object send job or to use a dedicated job.
Conversely, a consistently low difference between the last read entry and the current journal entry can indicate that more data groups may be able to share the object send job.
The optimal number of data groups sharing an object send process is unique to every environment. Determining the optimal number of data groups that can share an object send process in your environment may require incremental adjustments. Add or remove small numbers of data groups at a time to or from a shared object send process and monitor the impact on performance and throughput.
Factors that affect performance of a shared object send process include:
-
The number of data groups sharing the same object send job
-
The type of data replicated by the data groups sharing the object send job. It may be beneficial to share an object send process among data groups that replicate only IFS objects or only DLO objects.