Hi all, We have a Kafka Streams application which is showing high CPU usage. When profiling the application we see that many of the hotspots are related to RocksDB: flush, seek0, put iteratorCF and get methods.
We are using the default configuration for RocksDB. We read the documentation but would like to ask some pointers in case CPU is the bottleneck. We do a join and window step in our code. Commit interval is 1 second and we have caching enabled and the max cache size is 512MB. We have 5 instances (6 CPU, 30 GB RAM). Thank you!