A system cluster is comprised of two or more machines, known as nodes, tightly integrated through hardware and software to function together as one machine. In a cluster, redundant hardware and software are primarily put in place to enable fail over. Fail over is the movement of a cluster resource from one node in the cluster to another node. If something goes wrong on one node, or the node needs to be taken offline for maintenance, cluster resources can fail over to another node to provide continual access. Disk access is a key factor in proper fail over. If you use a clustered environment and wish to use Failover Clusters, you must have your database on a shared device. A shared device is a disk that is available to any node in the cluster. If a node in the cluster has an outage, the shared device is still recognized and available to the remaining nodes in the cluster, thus providing access to the database.
Failover Clusters provides a simple command-line interface, PROCLUSTER, to your operating system's clustering software. The Clusters interface is easy to use, and is the same regardless of the hardware or software platform. This simplifies administration of an OpenEdge database in a clustered environment. Because Clusters integrates your database into your cluster manager software, cluster resource administration and fail over mechanisms are enabled for your database.
Clusters does not replace OS-specific cluster management software. In fact, Clusters requires that the OS cluster management software and hardware be properly configured. See Relatedsoftware and hardware for your specific OS software and hardware requirements.
The operating system integrates specific components of its clustering technology to monitor the system resources' state and fail over the resource to another node if the primary node is not accessible. Clusters tightly integrates with the cluster management software so that OpenEdge is properly defined as a cluster resource and fails over during planned or unplanned outages. A planned outage might be a hardware or software upgrade. An unplanned outage might be a system crash. With Clusters, you can decide ahead of time how clustered resources will behave during fail over. Failover Clusters eliminates unnecessary downtime and provides continuity of behavior in the cluster even when the database administrator managing the cluster is not the one who set it up.
Clusters relies on the operating system cluster management for node selection and ordering. The node the cluster chooses to fail over to is a default preference Clusters accepts from the cluster manager. If you want to tailor this, you must use your operating system cluster management utilities. Clusters does not alter node ordering. All clustered systems provide a means for disallowing hosts from hosting a resource. Clusters does not provide a means to disallow a host; you must use your operating system cluster management utilities to do this.
