Collaborative Memory Management Assist (CMMA)

This z/VM support, in conjunction with the software CMMA exploitation in guest operating systems such as Linux® on System z™, allows the z/VM Control Program (CP) host and its guests to communicate attributes for specific 4 KB blocks of guest memory. This exchange of information can allow both the z/VM host and its guests to optimize their use and management of memory in the following ways:

• The CP knows when a Linux application releases storage and can select those pages for removal at a higher priority or reclaim the page frames without the overhead of paging-out their data content to expanded storage or disk.
• The CP recognizes clean disk cache pages, the contents of which Linux is able to reconstruct, allowing the CP to bypass paging-out the data contents when reclaiming the backing frames for these pages. If Linux or its application subsequently tries to refer to the discarded page, Linux is notified that the page has been discarded and can reread the contents from disk or otherwise reconstruct them.
• The guest further benefits from the hardware Host Page-Management Assist (HPMA), which was announced in the Hardware Announcement dated July 27, 2005. In conjunction with CMMA, HPMA allows the machine to supply fresh backing page frames for guest memory when the guest reuses a previously discarded page, eliminating the need for the z/VM hypervisor to intercept and resolve these host page faults.

To set the guests to use CMMA, we enabled the CP memassist function as follows:

CP SET MEMASSIST ON ALL