The DO managed Cache is great for simple applications, but applications that exceed 100k QPS should be making use of a cluster since Redis is single threaded. Think AWS ElastiCache.

The current setup does not allow a true clustering setup. You can add two standby nodes to read from however this is still not enough for some realtime applications that perform over 1mm QPS and requires connection pooling with a custom client. Its a work around to a much needed solution.

Also, please switch to Valkey 8 as it has better multithreaded support with IO