Generate datasets into benchmarks/data/.

# TPC-H (default: SCALE_FACTOR=1, PARTITIONS=16 - override by setting these environment variables)
./gen-tpch.sh

# TPC-DS (only SCALE_FACTOR=1 is supported)
./gen-tpcds.sh

After generating the data with the command above, the benchmarks can be run with:

WORKERS=0 ./benchmarks/run.sh --threads 2 --dataset tpch_sf1

• --threads: This is the physical threads that the Tokio runtime will use for executing the binary. It's recommended to set --threads to something small, like 2, for throttling each individual process running queries, and simulate how adding throttled workers can speed up the queries.
• --dataset: Dataset directory name under benchmarks/data/ (e.g. tpch_sf1, tpcds_sf1).

The same script is used for running distributed benchmarks:

WORKERS=8 ./benchmarks/run.sh --threads 2 --dataset tpch_sf1 --file-scan-config-bytes-per-partition 16777216

• WORKERS: Env variable that sets the amount of localhost workers used in the query.
• --threads: Sets the Tokio runtime threads for each individual worker and for the benchmarking binary.
• --dataset: Dataset directory name under benchmarks/data/.
• --file-scan-config-bytes-per-partition: How many bytes each partition is expected to scan. Lower values produce more partitions/tasks. Defaults to the engine default when unset.