Alibaba Cloud Hologres Tops TPC‑H 3TB Benchmark with Best Cost‑Performance

Alibaba Cloud Hologres recorded a score of 8,443,627 QphH@3000GB , placing it at the top of the official TPC‑H 3TB performance leaderboard and outpacing the second‑place system by roughly 54% . Its cost efficiency of 390.47 CNY/kQphH was better than every other system on the list, delivering both the highest performance and the best price‑performance ratio.

The result was independently audited by TPC certification auditor Doug Johnson of InfoSizing, and the full disclosure report (Result ID: 126062601) is publicly available on the TPC website for verification.

Historical Context

Hologres previously secured the global first place in the 2022 TPC‑H 30TB benchmark with 27,860,000 QphH, leading the runner‑up by 23%. Four years later, it reclaimed the top spot at the more enterprise‑relevant 3TB scale, expanding its lead to 54% while reducing unit cost.

Technical Optimizations (2022‑2026)

Over the past four years, the Hologres team addressed three core challenges:

1. Improving Compute Resource Utilization

Simply adding workers does not linearly increase performance due to data sharding and operator execution limits. Hologres rebuilt its query execution engine to stream massive data as columnar batches between operators, performing most calculations on lightweight‑encoded data and increasing operator parallelism without extra shuffle, thereby allocating more CPU time to actual data processing.

2. Reducing Invalid Data Processing

Complex queries incur overhead from data scanning, network exchange, and growing intermediate results. Hologres enhanced Runtime Filters and pushed down TopN operations, and its optimizer now learns from real query executions—using historical data distributions to correct cost estimates and automatically select better execution plans, allowing unnecessary data to exit the execution chain early.

3. Lowering Multi‑Load Interference

Real‑time warehouses handle concurrent workloads such as streaming writes, dashboard updates, scheduled reports, and ad‑hoc analysis. Hologres refined query thread‑pool isolation, compute‑group resource isolation, horizontal multi‑cluster scaling, and automatic load scheduling, giving different task types more independent execution space and maintaining stability under high concurrency.

The combined effect of these improvements is threefold: more compute resources are devoted to queries, less irrelevant data traverses the pipeline, and concurrent workloads experience reduced contention, enabling faster and cheaper real‑time analytics.

Impact on Real Business Scenarios

The 54% QphH advantage translates to lower latency and higher concurrency for typical multi‑table joins, multi‑dimensional aggregations, and reporting workloads in enterprise data platforms. The leading cost‑performance means this speed does not require prohibitive hardware expenditure. The architectural enhancements introduced since 2022 have become integral to Hologres, continuously supporting real‑time data warehousing, interactive analysis, and lake‑warehouse integration for thousands of customers.

Conclusion

From a 30TB benchmark lead of 23% in 2022 to a 3TB lead of 54% in 2026, Hologres has demonstrated repeatable, auditable performance gains and superior cost efficiency, confirming its technical roadmap of faster data updates, quicker query execution, and more stable performance under real‑world concurrent loads.