Database performance depends on many factors. There are many parameters user can tweak, but the outcome in complex setups is not always obvious. While we had a big cluster of crate instances
available, we faced the challenge to load huge amounts of data into it to initialize the distributed database. A series of experiments resulted in an awesome 4.8 mio records/rate of ingesting data into our cluster! To get there, we explored the effect of network connections, number of nodes, and shard configuration.