Configuration Rules for a Mixed HP 9000 / Integrity Serviceguard Cluster, March 2007
Table Of Contents
- Executive summary
- Version history
- Introduction
- Architectural rules and typical configurations
- Specific ISV information
- HP Superdome Hybrid Servers
- Typical configuration examples
- Transition to Integrity with HP Superdome hybrid servers and mixed clusters
- Adding one Integrity server to a two-node HP 9000 cluster
- Adding two HP Integrity servers to a two-node HP 9000 cluster
- Adding one critical application and two new Integrity nodes to existing four-node HP 9000 cluster
- Mixed clusters as transition aid from HP 9000 to Integrity in a multi-tier SAP environment
- HP 9000 to Integrity transition service utilizing HP Serviceguard cluster technology
- How to implement a mixed HP 9000 / Integrity HP Serviceguard cluster
- For more information
For configuration with applications that do not allow fail-over between heterogeneous cluster nodes
The examples below focus on how one can use mixed architecture clusters to their advantage even if
all or some of the applications don’t support failover between heterogeneous cluster nodes in the
mixed cluster. For the purpose of these examples, the applications labeled with “Database” are only
allowed to fail over between homogeneous nodes in a mixed cluster.
Adding one Integrity server to a two-node HP 9000 cluster
This example shows how one Integrity server is added to an existing two-node homogeneous HP
9000 cluster configuration that hosts a database that does not allow failover among heterogeneous
cluster nodes and an application server.
Original two-node HP 9000 cluster
Before the Integrity node is added to the homogeneous HP 9000 cluster, each node hosts a critical
application. The database and application are configured to fail over to the other node in case their
primary node goes down. The failover scenarios are:
• Node A fails.
– Node B runs both database and application servers.
• Node B fails.
– Node A runs both database and application servers.
Figure 4. Original two-node HP 9000 cluster
App Server
Database Server
Failover
PA
PA
Node A
Node B
OK
New three-node mixed HP 9000 / Integrity cluster
In this example, a new Integrity node has been added to the cluster for horizontal scaling on the
application server level. The failover scenarios are:
• Node A fails.
– No failover occurs; node C takes over application server load for all users.
– Optionally, the application server instance from node A fails over to node C if supported and
more preferred by the application.
• Node B fails.
– Node A runs the database and application servers (node C cannot act as failover node for
database service of node B).
– Optionally, the application server instance from node A can be halted or reduced in resource
entitlements through Workload Manager (WLM) if the remaining instance on node C can handle
the application server load alone.
18