z/OS DFSMS OAM Planning, Installation, and Storage Administration Guide for Tape Libraries
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


Scratch allocation assistance

z/OS DFSMS OAM Planning, Installation, and Storage Administration Guide for Tape Libraries
SC23-6867-00

With Release 2.0 of the TS7700 Virtualization Engine, additional support is provided for scratch allocations. This function is supported in a JES2-only environment at z/OS V1R10 and above with APAR OA32957, and in z/OS V2R1 with JES3 (refer to JES3 considerations for additional detail). If you have a hybrid configuration of interconnected TS7720 and TS7740 clusters, you might want to direct specific workloads to the TS7720 (for fast access) and other (archive-type) workloads to the TS7740. While this can be accomplished without scratch allocation assistance (SAA) through management class (copy) policies at the library, if the allocation request goes to the other cluster, IP (versus FICON) links are used by the library to send the data between the clusters (which is not as efficient). With scratch allocation assistance, you can establish policies at the library through a management class option that allows you to specify the clusters to which you want to direct the scratch allocation (candidate clusters) and only those clusters will be considered for the allocation request. If a management class is not specified through the ACS routines, “blanks” will be passed for the management class name indicating that the library’s default management class should be used. If none of the devices in the candidate clusters are currently available (perhaps they are all allocated), the existing MVS allocation options for device allocation recovery (WTOR | WAITHOLD |WAITNOH | CANCEL) are used. This support will also take other things into consideration (clusters being in service prep). If all of the candidate clusters are in service prep or there are no candidate clusters in the assigned management class, the allocation request will be honored as it is without scratch allocation assistance, with an online device being allocated from the available clusters. Copy policies can continue to be used that will direct one or more copies of the data to other clusters.

Scratch allocation assistance can be globally enabled or disabled through the LIBRARY REQUEST command (by default, this function is disabled). The DISPLAY SMS,LIBRARY command output indicates whether the composite library is enabled for scratch allocation assistance. With scratch allocation assistance enabled and requested through management class policies at the library (on an individual management class basis), the TS7700 will then surface a filtered (candidate) list of clusters to the host. Using this filtered list, MVS allocation will then randomly choose a device contained within this filtered list, the filtered lists of other eligible grid-configurations (composite libraries), or both. Refer to “Requesting information from the TS7700 Virtualization Engine for information on the LIBRARY REQUEST command. For additional information on scratch allocation assistance, refer to the TS7700 Information Center.

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014