Performance: Spark SQL vs. Databricks SQL (Serverless with Photon)

Yes, there is a notable performance difference between standard Spark SQL and Databricks SQL on Photon, especially when using Serverless SQL:

1. Databricks SQL with Photon (especially serverless) is heavily optimized for BI/analytical workloads. Photon is a vectorized query engine written in C++, designed to outperform JVM-based Spark SQL, particularly on complex joins, aggregations, and columnar scans. Printer support number
2. Benchmarks (including internal Databricks benchmarks and TPC-DS-style tests) often show 2x to 12x performance improvement for SQL workloads on Photon vs. traditional Spark SQL. These numbers, of course, vary by workload complexity, data size, and structure.
3. Serverless SQL adds auto-scaling, caching, and query optimization enhancements — often resulting in lower latency and better concurrency out of the box, without cluster management overhead.