2025 in Review: Databricks SQL, faster for every workload

Summary

In 2025, Databricks SQL achieved significant performance enhancements, delivering up to 40% faster execution across various workloads such as BI, ETL, and spatial analytics. These improvements are automatically applied without the need for manual tuning or query rewrites, making it easier for data teams to manage increasing data volumes and user concurrency. Key features include the introduction of Predictive Query Execution and Photon Vectorized Shuffle, which optimize query performance by default. Additionally, enhancements in Unity Catalog and Delta Sharing have streamlined data governance and sharing, ensuring that performance remains high even as data complexity increases.

Key Learnings

• 1Databricks SQL's performance improvements are driven by engine-level optimizations that require no manual configuration, allowing for seamless integration into existing workflows.
• 2Unity Catalog has significantly reduced end-to-end catalog latency, enhancing responsiveness in high-concurrency environments while maintaining strong governance.
• 3Delta Sharing now offers performance comparable to native tables, facilitating efficient cross-organization analytics without sacrificing speed.
• 4The introduction of Zstandard compression as the default storage format provides substantial cost savings while maintaining query performance.
• 5Geospatial analytics capabilities have been enhanced, allowing for complex queries to run significantly faster without the need for specialized systems.

Who Should Read This

Senior Data Engineers optimizing performance in large-scale analytics environments

Test Your Knowledge

Topics