MVThsm: Off-Host HSM Recycle

MVThsm: Off-Host HSM Recycle

ML2 without the Tradeoffs

HSM users have a long history of balancing mainframe resources against workload demands. In the case of DASD storage, ML2 offered to free up precious disk capacity in exchange for less expensive tape capacity. The catch was that HSM consumed expensive MSUs for data movement and Catalog Data Set (CDS) updates in order to reclaim tape capacity as data sets expired.

MVThsm shifts capacity optimization, normally achieved by the HSM recycle process, to the MVT tape infrastructure itself without mainframe CPU cycles for tape reads/writes or HSM CDS updates.

an efficient approach that continuously improves

MVThsm uses a lightweight started task to collect a list of expired data sets at customizable intervals. After that, the process of reclaiming ML2 tape capacity happens entirely on the MVT, with all tape reads and writes being performed without FICON I/O.

Tapes with expired data sets are transparently rewritten, replacing expired 16 kilobyte blocks with 16 byte blocks, keeping the same VOLSER and block IDs and reducing the size of expired data sets by 99.9%. As a result, the cloned tapes retain all of the location information for the remaining valid data sets, allowing the HSM CDS to continue using its existing references and skipping the need for the CPU-intensive process of updating the CDS.

MVThsm optimizes tape capacity without mainframe MSUs

MVThsm creates space efficient clones of ML2 tapes by pruning expired data sets without mainframe MSUs or the need to update the HSM Catalog Data Set.

Each time an ML2 tape is optimized by MVThsm, the tapes get smaller in size which, in turn, reduces the time for the cloning process and for replicating the tape to DR and thus continously improving performance.

By off-hosting capacity optimization, the process can be run more frequently, enabling existing MVT users to postpone tape storage upgrades and allowing new MVT users to fit existing workloads into smaller configurations.