[Feature Request / Performance]DirectQuery table is extremely slow when just draggin

[ysc625298945]

Dear Power BI Team,

I’m facing a critical performance issue when using Power BI Desktop + DirectQuery (connected to Hologres/PostgreSQL-compatible) for ad-hoc analysis. The problem is NOT on the database side, but caused by Power BI auto-generated DAX and Formula Engine (FE).

Steps to Reproduce (No custom DAX at all)

1. Use DirectQuery over a data warehouse table.
2. Simply drag dimension fields into a Table visual: Date, Region, Warehouse, Delivery Method, Express, Province, City, etc.
3. No measures, no complex calculations – just viewing details.

Issue

• Refresh time: ~58 seconds
• UI freezes completely
• Database SQL execution time: only ~1 seconds (returns 130k rows)
• 99% time spent on Power BI FE, not query

Root Cause (Proven by DAX Studio & ServerTimings)

1. Power BI auto-generates heavy DAX:
   • SUMMARIZECOLUMNS + ROLLUPADDISSUBTOTAL (forced subtotals/grand total)
   • Full sorting across many fields + TOPN(1000001)
2. DirectQuery pushes data warehouse-level aggregation to the single-threaded frontend FE
3. Even with Totals turned off, the query remains too heavy for ad-hoc usage

User Impact (Common enterprise scenario)

• Business users drag fields flexibly for ad-hoc analysis
• No DAX, no modeling – just simple tables
• DirectQuery must support lightning-fast detail queries

Suggested Optimizations

1. Optimize auto-generated DAX for detail tables
   • Do NOT generate ROLLUP/ISSBTOTOTAL by default
   • Reduce unnecessary sorting & nesting
2. Lightweight DirectQuery for details
   • Push down simple SELECT instead of heavy grouping
   • Add pagination / row limit by default
3. Isolate performance
   • Separate detail table rendering from heavy FE computation

Please fix this issue so DirectQuery can support fast, ad-hoc, drag-and-drop analysis for real business users.Thank you!servertimingstable