Alibaba Cloud's real-time data warehouse Hologres has posted a score of 8,443,627 QphH@3000GB, taking the top spot on the official TPC-H 3,000GB performance leaderboard at tpc.org — roughly 54% ahead of the second-place system. Hologres also delivered the best price/performance of any system on the leaderboard, at 390.47 CNY per kQphH (roughly US$58). First in both performance and value.
The result was independently audited by Doug Johnson of InfoSizing, a TPC-certified auditor. The Full Disclosure Report is published on the TPC website, where anyone can review and verify it (Result ID: 126062601).
Figure: TPC-H 3,000GB Performance Results page, captured July 2026.
This is not the first time Hologres has claimed the top spot. Back in 2022, Hologres won the TPC-H 30TB test with 27,860,000 QphH, finishing 23% ahead of the runner-up. Four years later, Hologres has reclaimed the No. 1 position — this time on the 3TB scale — widening its lead from 23% to 54%.
The 3TB scale is the "sweet spot" closest to real enterprise deployments: the core analytics workloads of most enterprises fall within this data range, which is exactly why the 3TB leaderboard is the most fiercely contested. A 54% lead at this scale — achieved at a lower unit cost — means Hologres is not just faster; it is more efficient.
This No. 1 was not won by throwing hardware at the problem. On a lower unit-cost basis, Hologres extended its absolute performance lead to 54% — fast and economical at once. TPC-H is widely regarded as the gold standard for data analytics benchmarking. For Hologres, the significance of this result goes beyond a bigger number: it provides a public, standardized performance benchmark that anyone can audit and verify.
Over the past four years, the Hologres team has tackled three core challenges: how to make distributed compute resources genuinely usable by queries, how to cut out wasteful data processing during complex SQL execution, and how to keep the system stable while real-time writes and high-concurrency queries happen at the same time.
In distributed query processing, adding workers does not automatically translate into proportional speedups — parallelism can be limited by data sharding and by how operators execute. To address this, Hologres rebuilt its query execution engine from the ground up: data now flows between operators in columnar batches, much of the computation runs directly on lightweight-encoded data, and operator parallelism is further increased without introducing extra shuffles — so more CPU time goes into actual computation.
Much of the cost of a complex query comes not from computation itself, but from data scans, network shuffles, and ever-growing intermediate results. Hologres has continuously enhanced runtime filters and TopN pushdown, and its optimizer learns from queries the system has actually executed — using real historical data distributions to correct cost estimates and automatically pick better execution plans for complex queries. The fastest data processing is removing the data you don't need from the execution path as early as possible.
A real-world real-time data warehouse never runs just one query: real-time writes, operational dashboards, scheduled reports, and ad-hoc analytics all compete for CPU, memory, and I/O within the same time window. Hologres has steadily improved query thread-pool isolation, compute-group resource isolation, multi-cluster horizontal scaling, and automatic workload scheduling — giving each workload a more independent execution space and keeping performance stable even under high concurrency.
This throughline comes down to three things: putting more compute resources to work on queries, letting less data reach downstream operators, and cutting wasteful waits across workloads. All of it was achieved on the same general-purpose engine that simultaneously powers real-time writes, high-concurrency point queries, and lakehouse workloads — Hologres was never an engine built just to chase benchmark scores.
These capabilities don't serve a fixed set of benchmark SQL; they serve the real requests that hit enterprise data platforms every day. A 54% QphH lead means the same multi-table joins, multi-dimensional aggregations, and reporting analytics run on Hologres with lower latency and higher concurrency.
And the No. 1 price/performance means that speed does not have to be bought at a steep price.
The improvements made over the past four years — across the execution engine, the optimizer, and the concurrency architecture — are already part of Hologres' product capabilities, continuously serving real-time data warehousing, interactive analytics, and lakehouse scenarios.
From 30TB in 2022 to 3TB in 2026, and from a 23% lead to a 54% lead, Hologres has twice proven its technical strength in complex analytics performance and price/performance — with results that stand up to audit. Leaderboard records will keep being broken. This milestone is both a checkpoint and a continuation of a long-term technical roadmap: fresher data, faster queries, and more stable performance in real concurrent environments.
Today, Alibaba Cloud Hologres serves more than a thousand enterprises, including 37 Interactive Entertainment, Muyuan Group, CaoCao Mobility, Happy Elements, ZhongAn Insurance, Leapmotor, and Semir.
Learn more about Hologres: https://www.alibabacloud.com/en/product/hologres
TPC, TPC Benchmark, and TPC-H are trademarks of the Transaction Processing Performance Council.
27 posts | 0 followers
FollowApsaraDB - August 31, 2021
ApsaraDB - August 14, 2026
ApsaraDB - July 14, 2025
Alibaba Cloud Community - November 3, 2022
Alibaba Cloud Community - April 17, 2026
Alibaba Cloud Community - July 8, 2026
27 posts | 0 followers
Follow
Hologres
A real-time data warehouse for serving and analytics which is compatible with PostgreSQL.
Learn More
Big Data Consulting for Data Technology Solution
Alibaba Cloud provides big data consulting services to help enterprises leverage advanced data technology.
Learn More
Big Data Consulting Services for Retail Solution
Alibaba Cloud experts provide retailers with a lightweight and customized big data consulting service to help you assess your big data maturity and plan your big data journey.
Learn More
ApsaraDB for HBase
ApsaraDB for HBase is a NoSQL database engine that is highly optimized and 100% compatible with the community edition of HBase.
Learn MoreMore Posts by Alibaba Cloud Big Data and AI
