HP 3000 Manuals HP 3000 Manuals
Introducing SPU Switchover/XL
by Rakesh Patel
Commercial Systems Division
The Hewlett-Packard high availability architecture is extended to include
a new product, SPU Switchover/XL (HP36378A). SPU Switchover/XL features
the ability of an SPU to take over the processing of information from
another SPU when so requested. Such a request could arise out of a
detection of the failure of an SPU, a need for achieving higher level of
processing on an SPU or a need to bring down an SPU for its maintenance.
In order to take over processing from another SPU, an SPU requires access
to the data residing on volume sets that normally belongs to the other
SPU. These volume sets are the switchable volume sets. Only one SPU has
continuous access to any particular switchable volume set, so that there
is no possibility of data corruption due to both systems having physical
connections to the disks. While it is obvious that SPU Switchover/XL
increases the data availability, we recommend using mirrored disks
(Mirrored Disk/XL HP30349) for the switchable volume set.
SPU SWITCHOVER/XL FEATURES
SPU Switchover/XL provides the following features:
* Switch over of data from one SPU to another.
* Recovery from hardware failures.
* Configurable switch over environment.
* Switch back of original processing from the alternate
* Supported on all MPE XL hardware platforms (3.0 and later)
supporting fiber links.
Switch Over of Data From One SPU to Another
SPU Switchover/XL provides increased data availability by using an
alternate SPU to take over the processing in case of system failure or
extended periods of maintenance. The data that needs to be highly
available should be made to reside on switchable volume sets.
Alternatively, volume sets that contain critical application data should
be configured as switchable volume sets. For the latter, each volume set
must be comprised of Fiber Link disk drives (for example, HP 7937FL). A
CI command allows a system to take over the switchable volume sets that
are home to the other system. If the latter system is up during this
command, the switch over process is carried out cooperatively with that
system.
When a system is switched over any user on that system will be logged
off. This is a result of either a system failure or being logged on to
the switchable volume sets.
Recovery From Hardware Failures
SPU Switchover/XL features the ability of one system to detect the
possibility of hardware (SPU) failure on the other system. This it does
by continuously monitoring the health of the other system by means of
heart beat between the systems. When a system detects the possibility of
a failure of the other system, a message will be displayed on its console
to alert the operator for the event. If the other SPU is indeed down
because of hardware failures, the operator may then decide to switch over
the critical processing to the live (alternate) system. In order to do
so, switchable volume sets are first switched over to the alternate
system, and then the applications are restarted, and users are logged on.
Configurable Switch Over Environment
The switch over environment is configurable through SYSGEN. A new first
level command presents the user with a list of SPU Switchover/XL related
set of commands. A user normally needs to configure only on one system,
and consistency of the configurations on both systems is automatically
maintained.
Switch Back of Original Processing From the Alternate SPU
The original state of processing can be restored (assuming switch over
was previously done) by switching back the switchable volume sets to
their home system. This is accomplished by a new CI command.
Transparent to Applications
SPU Switchover/XL is transparent to application design. No modification
of current applications is required to capture the full benefit of SPU
Switchover/XL.
Supported on All MPE XL Platforms
SPU Switchover/XL is supported on all MPE XL hardware platforms (XL
Release 3.0 or later) and requires disk drives equipped with fiber optic
links for switchable volume sets.
For further information, please refer to the related article in the
"Technical Articles" section in this Communicator.