Service interruption can occur due to many causes, from hardware failures to software problems on the hosted systems; or simply to allow for maintainance on the systems.

Load balancing between nodes can be obtained by configuring services in a flexible manner, so that they can be distributed between cluster nodes. This way, the problem of hot standby hosts that remain unused can be avoided.

Clusters usually require shared storage through shared SCSI cables or Fibre Channel. In order to safeguard data, storage is usually based around a RAID controller that handles the hard disks and the data within. This storage is partitioned in a way that each node can see all configured slices. Each slice is assigned to a single service. Since nodes share the resource, when they migrate services between them (TakeOver), they take control of each slice assigned to the migrated service.

Should a service become unavailable, day or night, KeyCluster users can count on our software for automatic service recovery (TakeOver), without having to depend on the constant presence of tecnical personnel or spare parts.

In order to be sure that the service will be restored, following common clustering filosophy, it is necessary to remove every Single Point of Failure, meaning each single resource that could render the cluster ineffectove in case of malfunctioning.