The above video shows a transactionnal queue replicated in real-time with the SafeKit data software replication and interrupted by a server failure. You will understand the difference between synchronous replication versus asynchronous replication. And you will see that on server failure, there is no data loss with a synchronous replication, while there is data loss with an asynchronous replication.
In the video, the Microsoft Message Queuing application is running on the primary server and is updating a transactionnal message queue. We first disconnect the replication network and you will see that the transaction counter stops because the replicated I/O cannot be acknowledged. Then, we simulate a server failure with a hardware reset of the primary server and you will see the automatic restart of the Microsoft message queuing application with no transaction loss thanks to the synchronous replication feature of the SafeKit data replication software.
On the same demonstration with asynchronous replication, all transactions made during the network failure are lost. If you consider that transactions are reservations in a aircraft, it means that you loose reservations after the server failure and passengers will have reservations for the same place. That’s why the SafeKit high availability software implements synchronous replication for application reliability.
If you want to make the right choice when choosing a data replication software solution, you can read this HA guide (High Availability guide). And you will find more information on synchronous replication vs asynchronous replication here…