---
title: "Introducing CostBench: an open benchmark for data warehouse cost-performance"
description: "Traditional data warehouse benchmarks evaluate raw execution speed without accounting for vendor-specific compute pricing. CostBench addresses this operational limitation by introducing an open framework designed to measure performance-per-dollar across cloud platforms. The initial benchmark release assesses read-side query efficiency across ClickHouse Cloud, Snowflake, Databricks, BigQuery, and Amazon Redshift using 43 production-derived analytical queries executed against an anonymized dataset. To normalize comparisons across varied billing metrics like credits, DBUs, slot-seconds, compute units, and RPUs, CostBench maps runtime execution directly to actual compute billing models. In this evaluation, ClickHouse Cloud remains in the low-cost, high-performance quadrant as data scales, outperforming its nearest competitor by 23x in cost-performance."
---

# Introducing CostBench: an open benchmark for data warehouse cost-performance

[Clickhouse](https://yomu.fyi/company/clickhouse) · Tom Schreiber, Lionel Palacin · May 27, 2026

**Type:** Benchmark

## Summary

Traditional data warehouse benchmarks evaluate raw execution speed without accounting for vendor-specific compute pricing. CostBench addresses this operational limitation by introducing an open framework designed to measure performance-per-dollar across cloud platforms. The initial benchmark release assesses read-side query efficiency across ClickHouse Cloud, Snowflake, Databricks, BigQuery, and Amazon Redshift using 43 production-derived analytical queries executed against an anonymized dataset. To normalize comparisons across varied billing metrics like credits, DBUs, slot-seconds, compute units, and RPUs, CostBench maps runtime execution directly to actual compute billing models. In this evaluation, ClickHouse Cloud remains in the low-cost, high-performance quadrant as data scales, outperforming its nearest competitor by 23x in cost-performance.

## Context

Comparing data warehouse performance is challenging because platforms expose costs through disparate billing metrics such as credits, DBUs, slot-seconds, compute units, and RPUs. Furthermore, agentic analytics and continuous data ingestion place severe cost and latency pressures on both write-side preparation and read-side query workloads.

## Approach / What changed

CostBench establishes an open benchmark that normalizes performance-per-dollar across ClickHouse Cloud, Snowflake, Databricks, BigQuery, and Redshift. It evaluates systems using 43 production-derived analytical queries over an anonymized dataset, mapping execution times to vendor billing models while open-sourcing all scripts, configurations, and raw results.

## Takeaways

- Speed comparisons alone are incomplete for cloud data warehouses because distinct billing units mask the actual compute costs required to complete workloads.
- CostBench evaluates cost-performance across two dimensions: read-side query execution and write-side ingestion efficiency.
- In the read-side evaluation across five cloud data platforms, ClickHouse Cloud outpaced its nearest competitor by 23x in cost-performance as data scaled.

**Tags:** [Open Source](https://yomu.fyi/topic/open-source), [Performance](https://yomu.fyi/topic/performance), [Scalability](https://yomu.fyi/topic/scalability)

- Source: [Clickhouse](https://clickhouse.com/blog/costbench-data-warehouse-cost-performance)
- Source URL: https://clickhouse.com/blog/costbench-data-warehouse-cost-performance
- Ingested by Yomu: 2026-08-28T04:01:05.259Z

[Read original post](https://clickhouse.com/blog/costbench-data-warehouse-cost-performance)