We are offering two types of redundant linux server configurations: active-active and active-passive. The basic netwok diagram for both configurations look like this:
Active-active configuration provides two servers that both share tasks and load. The way they share load is by using one server for one type of task and another server for another type of task, for example email is received and processed on one and web pages are served on another one.
Active-passive configuration is simpler: one server is working, another one is in stand-by mode.
In both configurations Linux High Availability Software is used.
The basic principle of the redundant server operation is the same for both active-active and active-passive configurations.
Here is what happens when one active server fails:
- The remaining server senses the failure via heartbeat subsystem.
- Failed server is shut-off completely to prevent data corruption by the remaining server.
- Remaning server mounts the storage that was used by the failed server, starts services that were
performed by the failed server and takes over the failed server IP's.
- Remaining server notifies tech support and/or customers via email and/or paging of the event.
- the entire process takes from 1 second to 1 minute after the failure is detected,
depending on the number and complexity of the services to take over.