Server Cluster Configurations (Recommended)
PPM Servers are deployed in a server cluster configuration. Clustering enables you to run PPM on several parallel servers, or nodes. Server cluster configurations improve performance on systems that handle high transaction volumes or large numbers of concurrent users. In addition to handling higher user loads and providing greater scalability, the server cluster configuration supports load balancing and server failover features.
Because clustering distributes work load across different nodes, if any node fails, PPM is still accessible through other nodes in the cluster. You can continue to improve system performance by simply adding nodes to the cluster.
To leverage the clustering capabilities within PPM to support either background service isolation or the load-balanced user traffic across multiple nodes, you must configure the instance (collection of nodes) as a formal cluster.
Caution: To avoid problems with memory and performance, Micro Focus strongly recommends that you isolate background services from user traffic. For more information, see Services Isolation.
In a server cluster configuration, you can enable multiple nodes to run background services.
In a clustered environment with Java Message Service (JMS) and Quartz clustered scheduling, you can configure the following:
Clustered nodes run across multiple machines
Number of consumers (listeners) per node for each queue
Service failure rules per queue (that is, number of retries, log failures, server shut-downs)
Notification messages to be sent to each queue
Specific number of threads per node and cluster information for the scheduler
Scheduler time zone. (This may be required if the database and PPM Servers are located in different geographies.)
To handle large numbers of concurrent users, server cluster configurations use either an external Web server or a hardware-based load balancer to distribute user connections evenly across multiple nodes. If more than one node in a cluster is dedicated to running services, and one of these services nodes shuts down, activities such as email notifications and executions scheduled to run on that node are automatically transferred to another available services node. This server failover feature helps ensure that PPM system services remain operational.
Note: Any unsaved changes on a node that shuts down are lost and are not transferred to an available node. Users who log on to PPM after a node shuts down see only changes that were saved on that node.
A PPM server cluster contains one or more nodes and an Oracle database. The first node installed and configured is the primary node. The other server (assuming a two-server setup) is the secondary node. The two servers can act as peers in a load-balancing situation, or one can act as a backup machine for the other.
Note: A server cluster configuration can include Oracle RAC. If a database in a setup such as this goes down, the Oracle JDBC driver manages database connectivity.
You can implement server cluster configurations on a single machine or on multiple machines. To run multiple PPM Servers on a single machine, the machine's memory capacity and CPU usage must meet the same memory and CPU requirements for multiple servers. To run multiple servers on multiple machines, the servers must share a common file system for reports, execution logs, attachment files, and server configuration files. Although each machine can contain its own instance of the PPM application code, only a single copy is required for each machine, regardless of the number of servers running on that machine.
You can set up server clusters with an external Web server, or with a hardware load balancer. The following sections describe these two setups.