You can reclaim zero pages and release the appropriate pages if the status of all data in the pages in the V-VOL reads no logical records (referred to as zero page). The free capacity of a pool increases after the pages are released. If you stop the reclaim zero pages operation, the zero pages that have already been reclaimed cannot be restored.
Logically, there is no difference between a page with zero data and the area of a DP-VOL without a page allotted. Both are effectively identical. However, a page with zero data uses pool capacity, whereas a DP-VOL without a page allotted does not use pool capacity.
Zero pages can be reclaimed when all these conditions are satisfied:
- The DP-VOL is not used in conjunction with another storage system software product that does not support reclaiming zero pages.
See Using Dynamic Provisioning for Mainframe with other software products.
- LDEV formatting is not being performed on the DP-VOL.
- The DP-VOL is not blocked.
- The DP-VOL is associated with a pool.
- The pool associated with the DP-VOL is not blocked, or is full and blocked.
- The DP-VOL is not a TSE-VOL.
Pages that include control cylinders are not processed when releasing pages in a DP-VOL.
While releasing pages from a DP-VOL, performance of the host I/O to the DP-VOL might temporarily decrease due to scanning for non-zero data.
If you stop an operation to reclaim zero pages in mid-stream, the pages that have been released will remain as free pool capacity.
After an operation to reclaim zero pages, Dynamic Provisioning automatically balances usage levels among parity groups in the pool. This rebalancing is performed on parity groups related to the pool. If you do not want automatic balancing of the usage levels of parity groups, call customer support to change your configuration.
Dynamic Provisioning does not automatically balance the usage levels among parity groups if the cache memory is not redundant or if the pool usage reaches the threshold.
If all the tracks that belong to a page assigned to a DP-VOL have no records written, you can reclaim the page and return it to the pool's available capacity.
In these cases, an operation of the reclaim zero pages stops and DP-VOL pages might not be released:
- The pool-VOL accessed by the target DP-VOL is blocked.
- The pool associated with the target DP-VOL is blocked while the operation to reclaim zero pages is in progress.
- Cache memory failure occurs while the operation to reclaim zero pages is in progress.
- While the operation to reclaim zero pages is in progress, the initial copy operation of the TrueCopy pair, the Universal Replicator pair, the global-active device pair, or the ShadowImage pair is performed on the DP-VOL.
- The LUN path is defined between DP-VOL and the host group enabled with host mode option 97 while the operation to reclaim zero pages is in progress.
- The pool-VOL accessed by the target DP-VOL is blocked.
- The pool associated with the target DP-VOL is blocked while the operation to reclaim zero pages is in progress.
- Cache memory failure occurs while the operation to reclaim zero pages is in progress.
- While the operation to reclaim zero pages is in progress, the initial copy operation of the TrueCopy for Mainframe pair or the Universal Replicator for Mainframe pair is performed on the DP-VOL.
User Directed Space Release (releasing ESE-VOL pages)
For ESE-VOLs, in addition to reclaiming zero pages, you can release unused ESE-VOL pages in a pool by using mainframe host commands. The function of releasing ESE-VOL pages by using mainframe host commands is called User Directed Space Release. The User Directed Space Release function is supported by the DKCMAIN microcode version 90-08-41 or later.
By running the SPACEREL command of DFSMSdss or the INIT command of ICKDSF, which are running on IBM® z/OS®, on an ESE-VOL, you can release the unnecessary pages allocated to the extent of the deleted dataset to increase the pool capacity.
For ESE-VOLs in a TPF environment, you cannot release pages because page release is restricted so that the I/O performance is not affected. Therefore, use DP-VOLs instead of ESE-VOLs in a TPF environment.
If no ESE-VOLs exist on the storage system, you can disable the use of ESE-VOLs and the User Directed Space Release function.
- The use of ESE-VOLs and the User Directed Space Release function are disabled by default. To change the settings, see the raidcom modify mainframe_system_function command in the Command Control Interface Command Reference. When the DKCMAIN microcode version is A0-05-41-x0 or later, you can also change the settings using VSP One Block Administrator. For more information about changing the settings using VSP One Block Administrator, see the VSP One Block Administrator Guide.
- If the subscription limit is set for an ESE-VOL, do not release pages by using the User Directed Space Release function. Because the reserved pages are not released even if you perform page release, the free pool space does not increase.
- The pages entirely included in the command range are released, but the pages partially included in the command range are not released. Also, the pages containing control cylinders, which are reserved for every 3,339 cylinders, are not released.
- If you run a command for an area whose capacity exceeds 4 TB (the sum of the maximum capacity for emulation type 3390-A for four LDEVs), the storage system might end the page release processing so that the host I/O is not affected. After running the command, use the LISTDATA command to check the number of released pages. See the information about viewing ESE-VOL or pool information by the LISTDATA command of the IDCAMS utility.
- While maintenance work is performed by maintenance personnel, page release by User Directed Space Release might be interrupted so that the host I/O is not affected. In this case, wait for at least 1.5 hours after the maintenance work is completed, and then check the capacity of allocated pages.
- While the User Directed Space Release function is working, if an MP unit is blocked or the controller to which the MP unit belongs is blocked, page release is interrupted. If this occurs, a timeout might occur although the interruption of page release is not reported to the host. Note, however, that the processing of page release continues in the units that are not blocked in the storage system. After 1.5 hours has passed since the page release was interrupted, check the capacity of allocated pages.
- A page of a partial pack minidisk created from z/VM in an ESE-VOL is not released by using the User Directed Space Release function from z/OS®. If the page is released, an abnormal end with CC=0008 occurs. For more information about the minidisk, see the IBM® documentation.
Whether these commands can be accepted differs depending on the type of the target volume, the software that uses the volume, and the pair status. The following table shows whether the commands can be accepted on the volume. If you run these commands under the condition in which the commands cannot be accepted, an error is reported.
If you run a command from a port set to I-2107-TPF or a command that specifies an operation to be performed from a port set to I-2107-TPF to the target volume, the pages are not released.
| Condition | Are commands accepted? | Are pages released? | ||
|---|---|---|---|---|
| Software that uses the ESE-VOL | Pair status | |||
| ShadowImage for Mainframe | P-VOL | V-Split/SUSPVS SP-Pend/TRANS | Yes | Yes (after transition of pair status)1 |
| Other than above | Yes | Yes | ||
| S-VOL | - | No | No | |
| Compatible FlashCopy® V2 | Source volume | - | Yes | No2 |
| Target volume | - | Yes | No2 | |
| Compatible FlashCopy® SE | Source volume | - | Yes | No2 |
| Target volume | - | No | No | |
| TrueCopy for Mainframe | P-VOL | Duplex | Yes | Yes |
| Pending Duplex | No | No | ||
| Suspend | Yes | Yes | ||
| S-VOL | - | No | No | |
| Universal Replicator for Mainframe | P-VOL | Duplex | Yes | Yes |
| Pending Duplex | No | No | ||
| Suspend | Yes | Yes | ||
| S-VOL | - | No | No | |
| Concurrent Copy | P-VOL | - | No | No |
| S-VOL | - | Yes | Yes | |
| Compatible XRC | P-VOL | - | No | No |
| S-VOL | - | Yes | Yes | |
| FICON® Data Migration | Source volume | - | Yes | No2 |
| Target volume | - | Yes | No | |
| Volume Migration | Source volume | COPY/PSUS | Yes | Yes |
| Target volume | COPY/PSUS | No | No | |
| Volume Retention Manager | Protect attribute | - | No | No |
| Read Only attribute | - | Yes | Yes | |
| Read/Write attribute | - | Yes | Yes | |
| Does not cooperate with software | - | - | Yes | Yes |
Notes
|
||||