What's Happening?
ClickHouse Cloud has demonstrated significantly superior performance and cost-efficiency compared to Databricks in a real-time analytics benchmark. The comparison, which involved streaming 113.2 billion stock quotes and running continuous queries, revealed
that ClickHouse Cloud delivered 752 times better end-to-end performance per dollar. This advantage stems from ClickHouse Cloud's ability to handle data preparation, including columnar storage, data ordering, and pre-aggregation, within its engine during the insert path. In contrast, Databricks spreads these tasks across separate services for ingestion, clustering, and materialized view (MV) refresh, leading to higher costs and potential data lag. The benchmark specifically highlighted ClickHouse Cloud's 24.2 times lower preparation cost, 54.2 times lower normalized query cost, and 31 times lower accumulated query runtime across both drill-down and interactive aggregation workloads. The test simulated continuous data ingestion at a target rate of 1 million rows per second, with queries running concurrently on ten-minute and hourly schedules.
Why It's Important?
This performance disparity has significant implications for U.S. businesses and industries reliant on real-time data analytics, particularly in sectors like finance, e-commerce, and logistics. Companies using Databricks for real-time applications may face higher operational costs and slower data insights, potentially impacting decision-making speed and competitive advantage. The benchmark results suggest that adopting ClickHouse Cloud could lead to substantial cost savings and improved data freshness, enabling businesses to react more quickly to market changes, customer behavior, and operational events. For instance, a trading platform like Bullet, which migrated from Databricks to ClickHouse Cloud, experienced a reduction in query latency from 10-15 seconds to milliseconds and improved data freshness from hours to under 5 seconds, while handling 1,000 times more data at a comparable monthly cost. This shift eliminates the need for separate serving layers, simplifying architecture and reducing infrastructure overhead, which is crucial for maintaining agility and efficiency in data-intensive environments.
What's Next?
The findings from this benchmark could prompt businesses currently using Databricks for real-time analytics to re-evaluate their data infrastructure and consider alternatives like ClickHouse Cloud. Databricks' new real-time warehouse, Lakehouse//RT, which is currently in beta, was not included in this comparison. Future benchmarks, once Lakehouse//RT becomes generally available, will be crucial to see if Databricks can close the performance and cost gap. Companies will likely continue to prioritize solutions that offer both high performance and cost-efficiency in real-time data processing. The trend towards more integrated and efficient data preparation and querying within a single system, as demonstrated by ClickHouse Cloud, is likely to influence future developments in the real-time analytics market, pushing vendors to innovate and optimize their offerings to meet growing demands for instant insights.
Beyond the Headlines
The benchmark results underscore a broader industry shift towards specialized, high-performance databases optimized for real-time analytical workloads. While general-purpose data platforms like Databricks offer broad capabilities, they may not always provide the most efficient or cost-effective solution for specific, demanding use cases like continuous ingestion and immediate querying. The emphasis on 'performance-per-dollar' highlights the increasing pressure on businesses to maximize value from their data investments. This also points to a growing sophistication in how companies evaluate data solutions, moving beyond raw processing power to consider the total cost of ownership, including data preparation, query execution, and the freshness of insights. The ability to maintain current summaries during data ingestion, as ClickHouse Cloud does, is a critical differentiator, as it directly impacts the accuracy and timeliness of real-time analytics, which can have profound effects on business outcomes and strategic decision-making.
