Documentation for C-Werk 2.0.

Previous page Configuring Failover VMS  Ports used by the failover system Next page

A failover system automatically prevents data loss when one of the Servers in the system fails.

In a failover system, the Servers are combined into a logical structure—the cluster. 

The C-Werk Failover system has two types of configuration.

  1. The basic configuration allows system supervisors to permit launching C-Werk servers (nodes) on any Servers within the system. 

    Note

    When selecting a Server to transfer a node to, the supervisor tries to keep in balance the whole cluster's performance. If all Servers deliver more or less the same performance, the selection is performed randomly.

    If Servers significantly differ in their performance, the supervisor may launch several nodes on a more capable Server, and no nodes on a less capable one. 

  2. In the configuration with the specified backup Server, a node from the primary Server can be migrated only to the backup Server. After the primary Server is back online, the node is returned.

Node migration is automatic and takes no more than one minute.

Note

In a system counting 100 cameras, the node is transferred in less than one minute in both Failover System configuration types. All Servers within the system have identical specifications: Intel i5-7400 3GHz 4-core CPU, 16Gb RAM.

This section contains the following terms:

  • Server—a computer with Failover Server and Client configuration of C-Werk installed;
  • Node—an instance of the Server services. A Server can have multiple nodes running;
  • Cluster—logical grouping of Servers that allows migration of nodes between them. A cluster may encompass nodes from different Domains;
  • Supervisor—the service that monitors the status of nodes and their migration.

To implement Failover in your system, we strongly recommend that:

  • the system administrator should have full control over all communication channels and hardware that provides fault tolerance at all times;
  • you should build a cluster from servers in the same LAN;
  • use only the network archives, that are available from all servers in the cluster.

Example of a Failover system:

  • No labels