TPC-H
The TPC-H Benchmark, published by the Transaction Processing Performance Council (TPC), is one of the most authoritative database testing standards. It is widely recognized by both industry and academia as a decision support testing standard and an important reference for database selection.
After rigorous large-scale analytical performance and ACID transaction tests, AnalyticDB for PostgreSQL topped the TPC-H 30 TB performance list, surpassing the current valid records to become the world's number one in both performance and cost-effectiveness. It is the first Chinese product to top the list.
In the TPC-H 30,000 SF list, Alibaba Cloud AnalyticDB for PostgreSQL 6.0 achieved 5,057,263 QphH with a cost-effectiveness of 1.46 CNY/QphH. Compared with the second-place result at the same scale—1,278,277 QphH from Cisco UCS C480 M5 running Microsoft SQL Server 2019—the performance is approximately 4 times higher.
The TPC-H Benchmark consists of business-oriented ad hoc queries and concurrent data modifications, and evaluates the transactional and analytical capabilities of databases.
The main challenges of this test are as follows:
30 TB large dataset: The 30 TB dataset used in this test contains a largest table of 180 billion rows, posing a huge challenge to data import, storage, and computing performance.
Complex association analysis: Multi-table joins, correlated queries, multi-dimensional data filtering, and high-precision numerical calculations challenge not only the optimizer (how to decorrelate queries, select the optimal join order, and choose the table distribution mode) but also the implementation of the compute engine (memory management, code generation, and execution scheduling).
Large-scale real-time writes: Write and delete operations on hundreds of millions of records test the performance of transaction processing and pose a huge challenge to the column-oriented storage engine required by AP systems.
High-throughput concurrent reads and writes: At least 10 concurrent query streams are required on the 30 TB dataset, while refresh streams perform multiple rounds of large-scale data insert and delete operations. This tests the concurrency and ACID capabilities of the system.
Distributed transaction verification: The transaction capabilities of the database system are fully tested, covering atomicity, consistency, isolation, and durability (ACID) tests, including shutdown tests of coordinator nodes and compute nodes. This verifies the transaction capabilities of the database system and the high availability of the system under extreme conditions such as power outages.
Related documentation
The TPC-H test requires multiple machine restarts and places strict requirements on the comprehensive capabilities and management of the system. For the detailed test process, seeTPC-H.