Fujitsu PRIMEQUEST 440, System Design Manual

Page: 219 / 280

C122-B001-02EN 9-1
CHAPTER 9 Clustering
1
This chapter describes clustering and items required to configure a cluster system.
To perform the cluster operation of the PRIMEQUEST, the clustering software
PRIMECLUSTER is required. For details of PRIMECLUSTER, see the relevant
PRIMECLUSTER manuals.
z The basics of clustering
z PRIMECLUSTER
z Cluster Functions Provided in Linux Environments
9.1 The basics of clustering
Clustering is the grouping of multiple systems, each operating independently, to have
them function as one system. Each of the systems forming a cluster is called a node.
Even if one node fails, another node takes over the processing, thereby enabling job
processing to continue.
Clustering can increase the availability of the entire system.
Availability is an index for the reliability of a system. A system in which few failures
occur, which continues operation even if a partial error occurs, and which can be
restored to its original state quickly following the occurrence of a system failure is
called a system having high availability.
By implementing auto detection of hardware failures and errors in middleware and
applications, and autonomous control such as taking over of job processing, job
processing can be continued securely.
During a scheduled stop for periodical maintenance or to change the system
configuration, the work can be done without affecting operating resources.
In PRIMEQUEST series machines, one cabinet can be divided into multiple logical
servers by the partition function so as to implement a cluster system, where each
partition acts as one node.