More efficient high availability and resource utilization through manageability

7
Figure 3 shows how WLM can adjust the virtual partitions (vPars) on the secondary server.
The following scenario involves six servers, each with a single application running, and one
Serviceguard secondary server. The secondary server has seven virtual partitions configured to handle
failovers from any of the six servers. Normally, the secondary server has a low-priority task running in
vPar 6 while all the other vPars remain idle, waiting for a failover. WLM is configured and running on
each vPar, ready to enforce any SLOs for packages that fail over.
When App 3 fails over to vPar 2, WLM detects the active Serviceguard package and allocates
additional CPUs to the application. If additional packages were to fail over, WLM would evaluate the
resource allocation among the active packages, shifting whole CPUs among vPars to meet the
applications’ SLOs. When App 3 returns to its primary server, WLM will reduce the CPU allocation to
vPar 2, reflecting its reduced need for CPU resources. As a result, WLM allows for a much less
expensive HA solution: the failover environment for many active servers can be a single server. If only
one application fails over, there is no performance degradation. If multiple applications fail over, the
resources are allocated to the highest-priority applications first. If those resources are not needed by
the high-priority application, WLM makes them available to the lower-priority applications.
Figure 3. Failover to server using vPars