I’m sure this is an uncommon request, but what is the production-grade setup for a single node scylladb instance? The use case is high-throughput writing with durability (more than postgres can offer), but not the availability requirement of multi-node.
For example, do I use SimpleStrategy because there is no need for network topology based strategy, and simple gossip?
Also guessing you never need to repair?
You won’t get HA, but you already know that.
Even with one node, there is no good reason to use SimpleStrategy. SimpleStrategy will make it hard to move to multi-DC in the future, and it needs an advantage.
Ok so keep network topology strategy, but what about repair and gossip, are my assumptions correct?
Yes, you do not need repair and gossip, but there is no need to disable either or any other service actively.