For each RMMplex, create one control data set on the main, or most active, system in your complex.

If the volume where you place the control data set is system-managed, select a storage class name that has the GUARANTEED SPACE attribute, and substitute it in the example for the STORAGECLASS parameter in Figure 1. If it is not system-managed, remove the STORAGECLASS parameter.

If you are running DFSMSrmm on more than one system, decide which systems in an RMMplex will share the control data set using DASD sharing and which will share the control data set using DFSMSrmm client/server support for z/OS. Each standard system and server system needs to share the control data set using DASD sharing. The control data set must be cataloged in a user catalog shared by all the standard systems and server systems in the RMMplex.

Put the control data set on a different volume than the journal, which is also shared. Separating the two data sets optimizes data integrity, since the journal is a copy of changes made to the control data set.

If you plan to use DFSMSdss to back up the control data set, place the control data set on a concurrent copy capable volume.

To avoid a potential deadlock on the volume where the DFSMSrmm control data set is placed, you should consider the other data that you place on the volume. A deadlock can occur when a program other than DFSMSrmm reserves the volume where the control data set resides and requests DFSMSrmm for service. If the DFSMSrmm control data set requires additional extents because of the request, then a deadlock can occur.

DFSMSrmm uses RESERVE/RELEASE to control access to the control data set and ensure integrity. If your installation is using global resource serialization, see Step 4: Updating installation exits and Updating GRSRNLxx (optional).

Place the DFSMSrmm control data set and journal on the highest performing DASD in your installation. DFSMSrmm can benefit from features like caching and DASD fastwrite and support for concurrent copy. Consider using the storage class attribute AVAILABILITY=CONTINUOUS for the control data set. This does not remove the need for journaling in DFSMSrmm, however, as the journal is required when reconstructing the control data set from backups.