More efficient high availability and resource utilization through manageability

5
configuration file and can be made available when a package fails over. If multiple users or
applications within a workload are competing for resources, standard HP-UX resource management
determines the resource allocation.
WLM enables you to define workloads based on:
• nPartitions that use Instant Capacity
• HP-UX virtual partitions
• HP-UX Integrity Virtual Machines (Integrity VM) hosts
• Resource partitions, which can be:
– Whole-CPU: HP-UX processor sets (pSets)
– Sub-CPU: Fair Share Scheduler (FSS) groups
Why use HP-UX Workload Manager with software high
availability clustering?
Combining WLM and clustering software results in more efficient high availability, which includes:
• Lowering the costs of standby systems (or partitions)—In active/standby configurations, the standby
systems go unused until a failover occurs. Using HP TiCAP or PPU software can help reduce the
costs of standby systems because the applications allow you to pay for only the CPU resources that
are used. With WLM, you can minimize the costs of using these CPU resources by optimizing the
activation and deactivation of the resources.
• Reducing the performance impact of an application failover in active-active configurations—
Consider a two-system Serviceguard cluster in which each system is running a combination of
packaged applications in an active/active configuration. (An active/active configuration has an
active system or partition as the secondary server, rather than an inactive system or partition just
standing by waiting for a failover.) If a package fails over from one system (or partition) to another
system (or partition), WLM can automatically reallocate resources as needed. On the system where
the package was located originally, WLM recognizes the package is no longer active. WLM is then
free to assign the extra resources to the remaining workloads. Similarly, on the package’s new
system, WLM recognizes the activation of a package. The system’s resources are then redistributed,
based on SLO priorities. As a result, the high-priority packages can maintain their performance
objectives, even in a failover scenario.
• Reducing the cost of disaster tolerance in active/standby configurations and the performance
impact of application failover in active/active disaster-tolerant configurations—Serviceguard
clusters can be deployed in disaster-tolerant configurations using the Extended Cluster, Metrocluster,
or Continentalclusters products. The benefits discussed in the preceding paragraphs for
active/standby and active/active configurations carry over to the multiple-site disaster-tolerant
configurations.
• Increasing utilization of a system that receives failover packages—When a system becomes
unavailable and its packages fail over to other servers in the cluster, WLM can detect which
packages are active and in need of resources. However, if resources remain after satisfying
high-priority packages, WLM grants the surplus resources to lower priority packages. You can also
provide a minimum amount of resources to the lower priority packages to guarantee that these
packages are not deprived of sufficient resources in a failover scenario.
• Simplifying the configuration of failover scripts—If your current failover scripts shut down low-
priority packages to make resources available for high-priority packages that have failed over,
using WLM can allow you to keep all the packages active. WLM will automatically reduce the
resources available to the low-priority workloads to maintain performance objectives of the higher-
priority workloads. Alternatively, if the system has TiCAP 6 or later or PPU 4, 7, or later resources