Cost-Effective High-Availability Solutions with HP Instant Capacity on HP-UX

Instant capacity components and high availability
Once you have installed iCAP components on a server, you have a foundation for the additional
types of iCAP. Also, you immediately have these two cost-effective features:
• In the event of a processor core failure (low-priority machine check or high-priority machine check),
an iCAP core replaces the failed core automatically. That is, when the system reboots following a
core failure, the HP-UX diagnostic monitor dynamically replaces a failed core with any inactive
core. This capability means that you can provision any server with standby iCAP processors, which
can be activated in case of processor core failures. This is a specialized solution for certain types of
hardware failures, but it can be deployed in any Serviceguard configuration and provides cost
benefits because you never pay the full price for the standby processing capacity.
• You can adapt load and priority changes among partitions of a server by migrating usage rights for
the components (cores, cells, and memory). You do this simply by deactivating resources in one
partition (which frees up usage rights) and activating the same type of resource in another partition.
This enables you to move resources between an active partition and a standby partition on the
same server. Alternately, you can move resources from a lower priority partition on a standby
server to a higher priority partition in a failover situation. This is cost effective because you use the
load balancing capabilities of Instant Capacity to pay for fewer resources in a standby partition.
However, the activation and deactivation can only be applied to a node or partition that is
running—not to a node or partition that is inaccessible because of a failure condition. (This can be
useful for planned maintenance and software upgrade situations.) When migrating cores and cells
through this deactivation/activation sequence, cores can be migrated in seconds; however, cell
status changes generally require a reboot. (The exception: When using HP-UX 11i v3 from
September 2007 or later, you can make cell status changes without rebooting.)
In addition, the HP Global Workload Manager product can automate the migration of core usage
rights among partitions based on policies you define.
HP Serviceguard considerations
Once iCAP components have been purchased and installed, the replacement of failed processors by
iCAP processors is automatic. You do not need to do anything to enable this feature in any
Serviceguard configuration (or any other configuration).
However, if you want to move resources automatically from one partition to another in a Serviceguard
configuration of dynamically resizable partitions, you must modify the package control scripts that are
used to start up and shut down the package. In those scripts, use iCAP commands to deactivate cores
from the active partition and then activate cores in a standby partition. (Cell changes may require a
reboot depending on the platform, and therefore, they may not be suitable for script automation.)
When the active partition is down, you must provide other mechanisms for failover situations because
you cannot deactivate or release resources from a partition that is down—unless you are using
GiCAP, which is discussed later in this paper.
7