I have a 5 nodes cluster and use a Replication Factor of 3.
I’m using the Read and Write Consistency Level of Quorum.
How many nodes can be down before Writes or Reads requests start to fail?
You can check that as well as other scenarios specific to your use case using the Consistency Level Calculator.
If you want a better understanding of Availability and what happens in different scenarios the High Availability lesson on ScyllaDB University is a good place to start.