Documentation / Query Benchmark / Overview

DBLense Query Benchmark — User Guide

Measure and compare SQL query performance with controlled iterations, concurrency, cache modes, wait stats, execution plans, CI regression gates, and the Query Tuning Assistant — on one server or across two.

Overview

Query Benchmark proves whether a SQL change is faster, slower, or equivalent before you ship it. It runs your SQL under controlled conditions and reports duration (including P50 / P90 / P95 / P99), CPU, logical/physical reads, degree of parallelism (DOP), query memory grants, TempDB usage, session wait stats, and optional plan comparison.

DBLense Query Benchmark main window with connections sidebar, mode tabs, Cache & Stats, Query A editor, and Run Benchmark
  • Benchmark — One connection, one SQL text, measured iterations.
  • Compare Queries — Run Query A and Query B (different SQL and/or databases on the same server) side by side.
  • Dual Environment — Two connections (Env A / Env B) for cross-server or version upgrade tests.
  • Config Impact — Measure the effect of proposed MAXDOP, memory, cost threshold, and related settings.
  • Index Analysis — Lab-friendly before/after index candidate runs.
  • Saved connections sidebar — Add/edit servers once; pick them on the main page to auto-connect (no Connect button).
  • Scenarios & File menu — Save/Load named scenarios, Export (PDF/HTML/JSON/CSV/Markdown/Excel), and open History.
  • Parameterized workloads — Bind @CustomerId-style parameters with Fixed, Random, Range, CSV, or Dataset value sources.
  • CI / CD Pipeline — Headless CLI with report files, JUnit XML, baseline regression gates, and exit codes for GitHub Actions / Azure DevOps. In-app snippets under Help → CI/CD Pipeline Integration.
  • Query Tuning Assistant — On-demand analysis: local SQL/plan heuristics plus live catalog/DMV checks (indexes, stats, Query Store, blocking, config). No cloud AI — everything runs on your machine and SQL Server.

Naming uses Query A and Query B (neutral labels) so the same UI works for rewrite compare on one server and cross-server tests.

Modes & test kinds

Mode tabs (underlined in brand cyan) select the experiment shape. Within a run you can still choose Performance, Load, Concurrency, or Stress (stepped user ramp) from Advanced settings, which stay visible on the setup panel.

Compare Queries workspace tab Dual Environment workspace with second connection panel

Deep dives: CI/CD Pipeline · Query Tuning Assistant. Watch a short product recording on the Query Benchmark demo page.