Cell Pool Management for Enclave Control Blocks (ENCBs)

Enclave control blocks (ENCBs) obtain their storage from a shared cell pool located in the z/OS Common Storage Area—a limited and critical system resource used by various workloads. When an enclave is created via the enclave create macro, a cell is allocated from the pool. Conversely, when an enclave is deleted using the enclave delete macro, the cell is returned to the pool.

If the cell pool is expanded to support thousands of concurrent enclaves, it remains at that size even after those enclaves are deleted. The pool is not contracted; instead, the storage is retained and reused for future enclave creation requests. This design ensures efficient reuse but also means that the allocated storage remains in memory until the system is IPLed.

Because these cell pools reside in the z/OS Common Storage Area, prolonged retention of unused cells can lead to excessive consumption of this scarce resource. If not managed properly, ENCBs may monopolize common storage, potentially leading to system instability or failure.

To improve memory efficiency and reduce the risk of monopolizing z/OS Common Storage Area, we propose adopting a garbage collection-inspired approach—similar to Java's GC—for managing cell pool storage used by enclave control blocks (ENCBs). This would involve periodically scanning the cell pool for unused or idle cells and compressing or releasing them when appropriate, rather than waiting for an IPL to reclaim memory.