Our ZOS people want to migrate the database from MOD9 storage to MOD27 in an effort to make the system more manageable. They believe that this would also perhaps also improve performance on the database. I couldn’t find anything that would support this. Has anyone had any experience with this?
I don’t think Adabas does anything differently as a result of the change. (There is one Empower posting that refers to MOD9, and it states this.)
You might see better (or poorer) throughput by changing to larger block sizes (839x device types).
Ask the manufacturer if the drives provide faster transfer rates or improved throughput.
We made the move to mod-54 volumes for production, but only once we had EMC Mainframe Essentials brought in house which supports parallel access (PAV) which is necessary to keep the unit control blocks (UCBs) from becoming a bottleneck. Though z/OS builds in support for PAV, the logical definition of large capacity volumes would make the OS think these are physically one device, regardless of whether your disk storage is actually physically different (e.g. RAID). Making use of PAV to allow multiple UCBs to access mod 27 or mod 54 volumes in parallel eliminates the risk of increased IOSQ delays in using these large volumes.
In making use of PAV with EMC Mainframe Essentials, we did see an improvement compared with mod-9s without it through a reduction in IOSQ delays including to ADABAS volumes even with moving to these larger model volumes, but I attribute this to some excellent analysis by the storage team, mainframe platform team and EMC services working together to identify the best way to define alias UCBs that allows the most optimal concurrent I/O processing.
We also have reduced the number of times maintenance is required for adding volumes and simplified the maintenance process in dealing with fewer volumes. And of course it will take a lot longer to reach the maximum volumes per dataset limit (59).