z/VM single system image and live guest relocation support

z/VM 6.2 introduces Single System Image (SSI) support, which is a multi-system clustering technology. You can cluster up to 4 z/VM images by using Single System Image. SSI facilitates resource sharing among the members in the cluster. You can move an active Linux on System z guest to another z/VM system without an outage of the guest. This facility is called Live Guest Relocation (LGR) and is only supported for Linux on System z guests.

To understand the z/VM SSI and LGR concepts and capabilities, refer to An Introduction to z/VM Single System Image (SSI) and Live Guest Relocation (LGR) (SG24-8006).

If the requisite level of z/VM is installed on the source and destination systems, the Live Guest Relocation feature of z/VM can be used to relocate a z/VM Linux guest system. If the requisite level of System Automation for Multiplatforms is installed on the Linux guest system, relocating this guest system does not affect the state or operation of the System Automation for Multiplatforms cluster if standard (default) heartbeat settings are configured. For an application that is managed by System Automation for Multiplatforms, the process of relocation is a brief suspension of operations. Restart is not required for System Automation for Multiplatforms and the application.

Validate that the period of suspension during Live Guest Relocation does not cause unwanted cluster events. Unwanted cluster events occur if a configured number of heartbeats is missing from the node that experiences a suspend during Live Guest Relocation. If testing shows that the average period of suspension can cause too many heartbeats to be missed, the heartbeat settings should be relaxed for the time of Live Guest Relocation. To greatly minimize the chance of unwanted cluster events while a z/VM guest system is relocated, stop the peer domain forcefully before the relocation is initiated by using stoprpdomain -f. For example, without stopping the applications that are managed by the cluster services. Upon successful completion of the relocation, restart the peer domain by using the startrpdomain command.