More efficient high availability and resource utilization through manageability

25
WLM is configured with resource policies to manage the applications in the cluster. All servers have a
total of eight CPUs. Each server has four licensed active CPUs and four deactivated Instant Capacity
CPUs so that systems can be expanded as needed. These Instant Capacity CPUs are important
because WLM can transfer Instant Capacity licenses from an Instant Capacity CPU on one nPartition
to an Instant Capacity CPU on another nPartition, providing the flexibility to “virtually transfer”
resources from one hard partition to another without incurring additional costs.
Assume the workload on one of the nodes in the first data center increases such that its SLO is no
longer being met. WLM determines if the other node in the server has licensed active CPUs that are
not being used. If such CPUs are available, WLM will transfer a license for one of them to the node
with the SLO that is not being met. If, however, no such CPUs are available, WLM can activate
Temporary Instant Capacity if present. If this server fails, Serviceguard will migrate the application to
another node in the cluster. With WLM configured in that standby server as well, it can manage those
Instant Capacity resources as just described.
Now assume an outage in San Francisco causes the entire data center to fail. The second data center
is configured to run the mission-critical applications, and the data has been replicated across data
centers using Metrocluster CA XP. Metrocluster verifies data integrity at the second data center and
informs Serviceguard, which then brings up the applications in the second data center. WLM will
adjust resources accordingly on nodes in the second data center to ensure SLOs are still being met.
By combining the VSE components nPartitions, WLM, and Instant Capacity, along with Serviceguard
and Metrocluster CA XP software, a robust, highly available, and disaster-tolerant solution with an
optimal total cost of ownership is now available.
Conclusion
The HP cluster products provide you with great protection against unavailable software and
hardware. Combining that power with WLM automatic resource management abilities produces a
much more efficient HA environment, an environment in whcih you are in control of all your
computing resources, regardless of what happens.
Related information
The following list of references provides useful background information on related products and
topics:
• HP-UX Workload Manager—http://www.hp.com/go/wlm
•
HP-UX Workload Manager User’s Guide —http://www.docs.hp.com/hpux/netsys/
• HP-UX Instant Capacity—http://www.hp.com/go/utility
•
HP Instant Capacity User’s Guide —http://www.docs.hp.com/hpux/netsys/
• HP Serviceguard and disaster-tolerant solutions—http://www.hp.com/go/serviceguard
• HP Serviceguard and disaster-tolerant solutions documentation—
http://www.docs.hp.com/hpux/ha/