The Process of Synchronizing Redundant Servers

[sbeadle1]

📖 Home  Back  
The synchronization process is used to ensure the Standby server in a Main-Standby server arrangement accurately reflects the data on the Main server.

The complete synchronization process occurs whenever a Main server re-establishes a connection with a Standby server. For example, a Main server can lose its connection with a Standby server due to a disconnected network cable. When the network cable is reconnected, the synchronization process begins. A synchronization will also occur when Geo SCADA is first started and connects to a Main server that is already running.

A shorter synchronization process occurs continuously when a Main server has a connection to a Standby server.

The complete synchronization process occurs when a Main server first connects to a Standby server or when a Main server re-connects to a Standby server that has been disconnected. The complete synchronization process is called the 'first pass' and it consists of these steps:

• The Main server transfers the database structure to the Standby server. The database structure is the hierarchy of groups, objects (items) etc.

• Configuration CRC (Cyclic Redundancy Checks) values are calculated from the specific configuration details of the database objects. The Main server retrieves the configuration CRCs from the Standby server and compares them to the configuration CRCs on the Main server.

• Any configuration data on the Main server that has different CRCs to the configuration on the Standby server is transferred to the Standby server. This ensures that the configuration data is the same on both the Main server and Standby server.

• Non-configuration data such as point states, timestamps, quality values etc., for all database objects is transferred from the Main server to the Standby server.

• Alarm data and event data is transferred from the Main server to the Standby server.

• The Main server queries the historic data modification time of the historic files on the Standby server. On small to medium sized systems, this should take a matter of minutes. On systems with large amounts of historic data (the design limit is 6 billion records), the historic data query could take hours.

• Any historic files on the Main server that have different modification times to the historic files on the Standby server are transferred to the Standby server. This will include files that have been modified for events logged since the last synchronization as well as any new historic files that have been created.

• The Main server queries the modification time of the Event Journal data on the Standby server. On small to medium sized systems, this should take a matter of minutes. On systems with large numbers of event records, the Event Journal query could take hours.

• Any event records on the Main server that have different modification times to the event records on the Standby server are transferred to the Standby server. This will include files that have been modified for events logged since the last synchronization as well as any new files that have been created.

• The Main server transfers the historic files (generated reports) to the Standby server.

When the 'first pass' is complete, the Main server will continue to run shorter 'passes'. Each 'short pass' consists of:

• The Main server checks to see if its database structure has changed since the previous 'pass'. Any changes to the database structure are transferred to the Standby server. The database structure is the hierarchy of groups, objects (items) etc.

• The Main server checks to see if there have been any configuration changes since the previous 'pass'. Any configuration changes are transferred to the Standby server.

• The Main server checks to see if there have been any changes to the saved data or dynamic data since the previous 'pass'. Any saved data or dynamic data changes are transferred to the Standby server.

• The Main server checks to see if there have been any updates to the alarm or event data since the previous 'pass'. Any new alarm or event data is transferred to the Standby server.

• The Main server checks to see if any new historic files have been created on the Main server since the previous 'pass'. Any new historic files or modified historic files are transferred to the Standby server.

• The Main server checks to see if any new event records have been created on the Main server since the previous 'pass'. Any new event records are transferred to the Standby server.

• The Main server transfers the historic files (generated reports) to the Standby server.

The 'short passes' run continuously to ensure that the Standby server matches the Main server.

On a small to medium sized system, a typical 'short pass' over a LAN should take around 1 second. However, the size of the database, the available bandwidth and the quality of the connection between the Main and Standby servers can all affect the amount of time it takes for a 'short pass'.

The rate of synchronization is dependent on the Update Rate setting (defined in the server configuration). For more information on the Update Rate setting, please refer to the Set up a Hot-Standby Pair section of the Server Administration Guide in Help.

Go: Home Back