Hitachi

JP1 Version 12 JP1/Performance Management User's Guide


10.1.1 Overview of cluster systems

Cluster systems are used to link multiple servers and run them as a single system. Cluster systems can be generally classified in the following two types:

Note:

In this section, cluster system refers to an HA cluster system.

Organization of this subsection

(1) HA cluster systems

The purpose of HA cluster systems is to enhance the availability of the entire system. HA cluster systems are often used in application servers and database servers for mission-critical systems that require high availability.

HA cluster systems include a redundant server for each server that makes up the system. In an HA cluster system, if problems occur in a server that is executing a job, a different server that has been standing by will continue the processing of the job. This is called failover.

Of the server systems in a cluster system, the system that is executing jobs is called the executing node, and the system that is standing by ready to take over processing when a problem occurs on the executing node is called the standby node.

The following figure shows the flow of access when a problem occurs on the executing node.

Figure 10‒1: Flow of access when a problem occurs on the executing node of an HA cluster system

[Figure]

The software that controls the entire HA cluster system is called the cluster software. The cluster software always monitors the servers and, if a problem occurs, automatically switches the server for executing a job from that on the executing node to that on the standby node. Therefore, cluster systems are also called node switching systems.

(2) Load-balancing cluster systems

Load-balancing cluster systems balance the processing load across multiple servers. They are often used in systems that require high processing performance.

Load-balancing cluster systems place multiple servers in parallel to balance processes, keep the load on any single server low, and increase the processing performance of the entire system. Even if a problem occurs on a server, switching processes to a different node can enhance the availability of the system.

Figure 10‒2: Flow of access in a load-balancing cluster system

[Figure]

Supplemental information:

Examples of load-balancing cluster systems include systems that balance servers to receive requests, such as Web systems, and Oracle Real Application Cluster systems. In addition, business applications that run on load-balancing cluster systems require programs that can allocate processes to multiple nodes.