Hitachi

JP1 Version 12 JP1/Data Highway - Server Configuration and Administration Guide


C.1 Overview of a cluster system

A cluster system consists of multiple server systems that work together so that they can operate as a single system. If one server fails in a cluster, another can take over the task.

The cluster system is made of one host that processes tasks and another that is waiting so that it can take over in the event of failure. The former host is called an active node, and the latter, a standby node. If a failure occurs on the active node, the standby node takes over the processing to continue tasks. This takeover action is called failover.

A failover is performed based on a logical node, which is a logical host. Applications in a cluster system must run within a logical host environment in order to fail over to continue tasks. When the applications run on a logical host, they are independent of a physical server machine, allowing them to run on any server machine.

A logical host consists of three components: applications running as a service or daemon, shared disk, and a logical IP address. The applications, such as JP1/DH - Server services, store their data in the shared disk and use the logical IP address to communicate with other applications.

The following table lists and describes each component of the logical host.

Table C‒1: Components of a logical host

No.

Logical host component

Description

1

Service or daemon

JP1/DH - Server applications running in a cluster system. If an active logical host fails, a standby logical host starts the services or daemons with the same name and takes over.

2

Shared disk

Disk unit connected to both the active and standby systems. It contains information for failover operations, such as definitions and operating status. In the event of a failure in an active logical host, the standby logical host is connected to the shared disk as the active host.

3

Logical IP address

An IP address allocated to a running logical host. If the active node fails, then the same logical IP address is allocated to the standby node. Therefore, clients can continue access via the same IP address as if only one node is always running.