From your link: "This was achieved with 32 (out of a maximum 48) data nodes." The project being discussed is benchmarked on a single node, not a cluster. This is why they compare performance to RocksDB.
From your link: "This was achieved with 32 (out of a maximum 48) data nodes." The project being discussed is benchmarked on a single node, not a cluster. This is why they compare performance to RocksDB.