Consensus algorithms underly distributed databases like Cassandra, or distributed caches like Redis. (in fact Redis does use a version of Raft).

Simple use case, you're running a database across three servers, so that users can ping any one of the servers and it still works (fault tolerance).

User X pings server A to write something to a table. User Y pings server B to read from that table.

How do you make sure that User Y reads what User X wrote, so that User X and User Y are on the same page with what happened?

That's the consensus algo