ankushrayabhari / kushdb

Compilation-Based Execution Engine for a Database

KushDB

Compilation-based execution engine for a database

Execute TPC-H, JCC-H

0. Instructions will be for SF 1 but same holds for other scale factors.
1. Run dbgen to generate .tbl files.
   • For TPC-H, use https://github.com/electrum/tpch-dbgen.
   • For JCC-H, use https://github.com/ldbc/dbgen.JCC-H.
2. Move them to benchmark/tpch1/raw/.
3. Run bazel build //benchmark/tpch1:load && bazel-bin/benchmark/tpch1/load.
4. Run python benchmark/tpch1/monetdb_load.py.
5. Run python benchmark/tpch1/duckdb_load.py.

KushDB

To bench, run python benchmark/tpch1/bench.py > tpch1.csv.

MonetDB

To bench, run python benchmark/tpch1/monetdb_bench.py > monetdb_tpch1.csv.

DuckDB

To bench, run python benchmark/tpch1/duckdb_bench.py > duckdb_tpch1.csv.

Execute JOB

1. Download CSV Files from JOB here.
2. Make the following changes to the CSV files:

• Update line 268268 of person_info.csv to escape the \ in Canada\US
• Update line 2671662 of person_info.csv to escape the \ in (qv); \,
• Update line 2514451 of title.csv to escape the \ in \Frag'ile\

3. Run python benchmark/job/monetdb_load.py.
4. Run python benchmark/job/duckdb_load.py.

MonetDB

To bench, run python benchmark/job/monetdb_bench.py > monetdb_job.csv.

DuckDB

To bench, run python benchmark/job/duckdb_bench.py > duckdb_job.csv.