TPCDS

TPC-DS is a decision-support benchmark created by the Transaction Processing Performance Council (TPC). It is designed to evaluate the performance of database systems, data warehouses, and analytics platforms under a realistic, enterprise-scale workload that resembles retail decision support.

The benchmark uses a comprehensive data warehouse schema composed of fact and dimension tables that model a multi-channel retail scenario, including store, catalog, and web sales. The workload consists of a fixed set of SQL queries designed to exercise a broad range of analytic patterns—joins, aggregations, complex groupings, window functions, and subqueries—along with maintenance and data loading tasks. The goal is to reflect real-world decision-support tasks such as revenue analysis, product and customer analytics, and operational reporting.

Data for TPC-DS are generated deterministically via a scalable data generator. A configurable scale factor sizes the dataset, allowing benchmarks to range from small to very large data warehouses. This scalability enables testing across different hardware and cloud configurations while maintaining reproducibility.

The primary metric reported by TPC-DS is QphDS (Queries per hour for the TPC-DS workload). Results are presented for each scale factor and must be certified under TPC rules. Additional measurements often include data loading times and maintenance times, providing a complete view of both query performance and data-management overhead.

Historically, TPC-DS was introduced as a successor to TPC-H to provide a more realistic, decision-support-focused workload. It remains widely used for evaluating analytical database systems and cloud-based data warehouses, with results published by vendors and researchers to compare performance across platforms.