Common fault tolerance mechanisms: failover, failsafe, failfase, failback, forking, which are derived from Ali's definition.
- Failover failed automatic switch
When a failure occurs, retry other servers, usually for read operations (recommended). Retrying will introduce longer delays.
- Failfast fail fast
Only one call is made, and an error is reported immediately if it fails, which is usually used for non-idempotent write operations. If a machine is restarting, the call may fail.
- Failsafe
When an exception occurs, it is ignored directly, usually used for operations such as writing audit logs. Loss of call information can be used to monitor in production environment.
- Failback failure automatic recovery
The background records the failed request and resends it regularly. Typically used for message notification that the operation is unreliable and restarts are lost. Registry available for production environments.
- Forking calls multiple servers in parallel
As long as one is successful, it will return, which is usually used for read operations with high real-time requirements. More service resources need to be wasted.
- Broadcast
Broadcast calls, all provide one by one call, any one will report an error. It is usually used to update the local state of the provider at a slow speed, and an error will be reported if any one reports an error.