- Cloud Analytics Platform Offerings
- Benchmark Setup
- Benchmark Results
- Price Per Performance
- About the Authors: William McKnight & Jake Dolezal
- About Gigaom Research
Product Profile and Evaluation: Microsoft Azure SQL Data Warehouse and Amazon Redshift
Data-driven organizations rely on analytic databases to load, store, and analyze volumes of data at high speed to derive timely insights. This benchmark study focuses on the performance of cloud-enabled, enterprise-ready, relationally based, analytical workload solutions from Microsoft Azure SQL Data Warehouse and Amazon Redshift.
For our benchmark, we compare the performance of SQL Data Warehouse and Amazon Redshift using a workload derived from the well-recognized industry standard TPC Benchmark H (TPC-H). The intent of the benchmark’s design was to simulate a set of basic scenarios to answer fundamental business questions and report business outcomes relevant to many industry sectors. In our opinion as field practitioners, this benchmark is a fair representation of the majority of queries that are run on data warehouses, excluding the low-end basic queries and the high-end data science queries.
The benchmark tested the scalability of corporate-complex workloads in terms of data volume with 30TB of data. The testing was conducted using as similar a configuration as can be achieved across Azure and Amazon Web Services (AWS) offerings.
In our benchmark, all 22 queries of the TPC-H (with only necessary platform-specific syntax changes) were executed serially as a single set three (3) times (after a warmup). The individual times of each query were added, and the set with the fastest overall run of the set was chosen for comparison (in aggregate and for individual query comparisons).
The benchmark results were insightful in revealing the query execution performance of Azure SQL Data Warehouse and Redshift and some of the differentiators in the two products. Azure SQL DW outperformed Redshift in 56 of the 66 queries ran. Overall, the performance advantage was 1.67 times faster.
Please note these results are as of July 2018. Future enhancements to either platform could shift the results.