ClickHouse / ClickHouse/ClickBench

Some, but not all scripts report "load time" for stateless engines (time to download parquet and sync)

Open
#944 0 comments 1 reaction 0 assignees View on GitHub
Dominant language
Shell
Stars
1.1k
Forks
313
Avg merge
2h 43m
Merged PRs (30d)
598

Description

While reviewing recent DataFusion results, I found there is now a "load time" reported for DataFusion

For example, this [link](https://benchmark.clickhouse.com/#system=+drB|erDt|ti%20rud&type=-&machine=+ca4e&cluster_size=-&opensource=-&hardware=+c&tuned=+n&metric=combined&queries=-)

Image

Here is the source file that has 17:
https://github.com/ClickHouse/ClickBench/blob/a377499b950b17b2d96b79625375d57b18a36634/datafusion-partitioned/results/20260511/c6a.4xlarge.json#L10

It was added in https://github.com/ClickHouse/ClickBench/commit/6d5ee0a68573bbb481fdb8f799cc3348a9bb0634 which was part of a large refactoring in https://github.com/ClickHouse/ClickBench/pull/860

Image

I am not sure if you intended the load_time to include the download / fsync time for the raw parquet dataset, but if you did I would request that we measure it consistently across stateless engines

Contributor guide

No contributing guide indexed for this repository

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.