From 48a491ff2f98f04774512ba308c9786fe0b52d38 Mon Sep 17 00:00:00 2001 From: Geoffrey Yu Date: Fri, 26 Apr 2024 13:10:16 -0400 Subject: [PATCH] Check in table size stats for CH-BenCHmark --- src/brad/planner/constants.yml | 16 ++++++++++++++++ tools/calibration/table_sizes.py | 4 ++-- 2 files changed, 18 insertions(+), 2 deletions(-) diff --git a/src/brad/planner/constants.yml b/src/brad/planner/constants.yml index 51d9a428..4c393d51 100644 --- a/src/brad/planner/constants.yml +++ b/src/brad/planner/constants.yml @@ -181,6 +181,22 @@ table_extract_bytes_per_row: movie_info: 29.57191 person_info: 133.458044 + # TPC-C Warehouses: 1740 + # Around ~120 GB of uncompressed data. + chbenchmark: + warehouse: 92.40747126436781 + item: 75.62581 + stock: 308.868974 + district: 98.52431034482758 + customer: 570.148704 + history: 65.51127 + orders: 40.134002 + new_order: 9.937048 + order_line: 68.538322 + region: 216.8 + nation: 185.03225806451613 + supplier: 194.728 + ### ### Models used to account for hardware/system load. ### diff --git a/tools/calibration/table_sizes.py b/tools/calibration/table_sizes.py index ca1c3bc4..0fce773d 100644 --- a/tools/calibration/table_sizes.py +++ b/tools/calibration/table_sizes.py @@ -30,7 +30,7 @@ def delete_s3_object(client, bucket: str, key: str) -> None: async def main_impl(args) -> None: - config = ConfigFile.load(args.config_file) + config = ConfigFile.load_from_physical_config(args.physical_config_file) assets = AssetManager(config) mgr = BlueprintManager(config, assets, args.schema_name) await mgr.load() @@ -121,7 +121,7 @@ def main(): "Run this after bootstrapping a schema to measure table sizing " "constants used by the blueprint planner." ) - parser.add_argument("--config-file", type=str, required=True) + parser.add_argument("--physical-config-file", type=str, required=True) parser.add_argument("--schema-name", type=str, required=True) parser.add_argument("--debug", action="store_true") # Unloading is slow - we do not need to unload the entire table to get a