HP StorageWorks EVA Cluster Extension Software Installation Guide (T3667-96044, April 2010)

• Linux: For SLE HA and RHCS data backup requirements, see the SLE HA and RHCS documentation.
CAUTION:
EVA Cluster Extension works with one server at each location, with a single I/O path between the
server and storage system, and a single link between storage systems. However, any configuration
with a single point of failure is not considered highly available or disaster tolerant. HP does not support
configurations with single points of failure; configurations with one system at each location are
supported only if there are no single points of failure.
How EVA Cluster Extension Software interacts with its
environment
This section describes the ways in which EVA Cluster Extension interacts with its in environment.
Failover function
EVA Cluster Extension Software requires cluster software to automatically fail over among systems on
a local site or between sites. Because the software manipulates the application startup process before
storage system vdisks are accessed, it must be integrated as the first resource (in the order of resources).
The software must be installed on all servers in the cluster that can run the application service.
How EVA Cluster Extension utilizes vdisks
To write to Continuous Access EVA paired vdisks, you must make sure the paired vdisks are in
read/write mode. DR group members (paired vdisks) are usually in read/write mode on the source
(primary) vdisk only. The destination (secondary) vdisk is in read-only mode.
• In case of a failover, EVA Cluster Extension Software automatically changes the direction of the
mirrored pair. A reboot of the server is not necessary.
• In case of a disaster, the storage system can have several different states for DR group members.
EVA Cluster Extension Software determines whether those vdisks can be accessed and made
available to the application service.
How EVA Cluster Extension stores information
EVA Cluster Extension stores information about the application environment in an internal object
database and uses HP SMI-S EVA to gather information about the state of the associated DR group
members. Information about the configured storage system environment and failover behavior is
transferred either directly by the cluster software or by gathering it from the user configuration file.
An internal object database provides the software with the format and default value of supported
parameters.
Vdisk states are stored in the internal object database, which uses a rule engine to process those DR
group member states. The rule engine matches current DR group member states and configuration
parameters with a defined rule, stores it in the database, and invokes predefined actions. Those
actions prepare the DR group member vdisks to be enabled for access or stop the application service
startup process if the matching rule requires a stop.
Preparing to install EVA Cluster Extension8