Benchmarking Apache Cassandra (40 Nodes) vs. ScyllaDB (4 Nodes)

I don’t understand why they haven’t used identical hardware, network latency and CPU context switching can be tricky to measure correctly.