Relief from storage shortages in group buffer pools
In Db2 11, data sharing members can automatically respond to storage shortages in group buffer pools with minimal impact to your applications. This new behavior is called the group-buffer-pool write-around protocol.
Before Db2 11, when a group buffer pool sustained a large amount of page-write activity, the coupling facility for the group buffer pool could quickly fill up with changed pages. As a result, applications could be slower, and in severe cases, pages could be placed in the LPL. These LPL pages could result in Db2 data outages.
In Db2 11, Db2 automatically detects large amounts of page-write activity and switches to the group-buffer-pool write-around protocol for those objects that have the most write activity. (Typically, the objects that have the most write activity are utility and batch jobs.) Under this new protocol, pages that are changed by write operations are written to the group buffer pool only if they are already cached there. Otherwise, Db2 writes these pages to disk. The other objects (objects that are using the group buffer pool but do not have the most write activity) are not affected by this new protocol. They are able to access the group buffer pool as usual. When the storage shortage is relieved, Db2 switches back to normal write activity for the group buffer pool.
Db2 automatically implements the group-buffer-pool write-around protocol when necessary. However, the following prerequisite software is required:
- z/OS® Version 1 Release 12 or Release 13 with the PTFS for OA37550 or z/OS Version 2 Release 1
- Coupling facility control code (CFCC) at CFLEVEL=17
To determine how many pages are written to disk by the group-buffer-pool write-around protocol, issue the DISPLAY GROUPBUFFERPOOL command with the MDETAIL option. In the output from this command, look at the WRITE-AROUND PAGES field in message DSNB777I.