In the client/server mode, using a server cluster provides fault tolerance. The system ensures that work is uninterrupted in the event of hardware or software failures in the server cluster.
User experience is therefore not impacted in any way by such events as working server failures, central server failures, and emergency or planned shutdowns of working processes or cluster managers.
If a user's connection to the cluster is physically lost (for example, if a cable is unplugged), the user can continue their regular work until the infobase connection is restored and their data will not be lost.
The following measures provide fault tolerance:
- cluster redundancy;
- working process redundancy;
- tolerance of connection losses.
Several clusters can form a redundancy group. Clusters belonging to the same redundancy group are automatically synchronized.
If an active cluster becomes unavailable, the next available cluster in the group is activated. When a cluster whose position in the group precedes that of the one currently active becomes available, it becomes active once automatic data synchronization is performed.
Working process redundancy
For each working process, you can specify the usage option: Use, Use as backup, or Do not use.
In the event of the emergency shutdown of a working process, the cluster starts one of the inactive backup processes and rebalances the workload.
Tolerance of connection losses
As it opens a separate session for each user, the cluster "remembers" the users connected and the state of the actions they are executing at the time the connection is lost.
If a physical connection is lost, the cluster will wait until the connection is restored. In most cases, after the connection is restored, the user can continue their work from the point where it was interrupted. Re-establishing the infobase connection is not required.