Skip to content
Databricks SQLChangelog · September 2026

System query history reached general availability, improving query monitoring across Databricks SQL.

Databricks marked system.query.history as generally available, giving administrators and developers a standardized system table to analyze warehouse workloads and audit execution metrics across accounts [1]. In analytical functions, Databricks SQL added the counter_diff window function to simplify calculating differences in cumulative counter values across time-ordered partitions [3]. Ecosystem tooling also advanced with release v1.12.5, which improved log record interpolation for Databricks SQL and updated metric view replacements with a backup-and-create pattern [15].

Query telemetry and warehouse cost governance drew substantial focus during the month. Practitioners shared operational queries that combined system tables with Databricks SQL alerts and the AI_FORECAST function to track warehouse consumption and project 30-day expenditures [4]. These monitoring patterns fed into broader community debates regarding cost attribution, with users calling for immediate cost visibility on job runs [5] and analyzing whether workload design rather than base pricing drove expenses for smaller teams [11, 12].

Practitioners also worked through runtime behaviors and compute selection. Developers addressed discrepancies where queries ran successfully inside the Databricks SQL interface but failed when submitted through JDBC connections [14]. Meanwhile, discussions focused on foundational architectural questions, including how SQL warehouses differ from classic compute [13, 16], experiments in routing SQL traffic through zero-shot decision models [2], and feature restrictions to weigh before migrating workloads to serverless options [6].

Everything cited

  1. [1]system.query.history is now Generally Available in Databricks community · 2026-09-30
  2. [2]Laya off the benchmark: can a zero-shot decision model route real SQL traffic? community · 2026-09-24
  3. [3]New Databricks SQL window fuction: counter_diff community · 2026-09-23
  4. [4]Top 5 System Table Queries for Understanding Your Databricks Costs news · 2026-09-22
  5. [5]What if Databricks showed the cost directly next to every job and run? community · 2026-09-21
  6. [6]Before you move a job to serverless: the limitation list is the real decision, not the startup time community · 2026-09-21
  7. [7]Databricks pipeline works in one workspace but fails in another — where would you start? community · 2026-09-20
  8. [8]Manager wants us to "use AI." Thinking about an AI-driven data testing framework for DevOps promotions. Sanity check? community · 2026-09-14
  9. [9]How do you parse an xml that's in string format? community · 2026-09-12
  10. [10]No more UNION ALL-ing all of your SDP pipeline event log tables for monitoring community · 2026-09-10
  11. [11]Databricks is too expensive for small teams" is usually a workload problem, not a platform problem community · 2026-09-10
  12. [12]Do small companies actually use Databricks? community · 2026-09-07
  13. [13]What is a Warehouse in Databricks SQL? community · 2026-09-04
  14. [14]Query works in Databricks SQL but fails through JDBC community · 2026-09-03
  15. [15]v1.12.5 release · 2026-09-01
  16. [16]What type of compute do you guys use in Databricks? community · 2026-09-01