Skip to content

Benchmark Dashboard

The benchmark dashboard is a self-contained HTML report showing QueryTurbo cache performance. It is generated from a point-in-time snapshot of the current process's SQLCompilationCache state.

Generating the Report

python manage.py query_doctor_report

By default, the report is written to query_doctor_report.html in the current directory.

Options

Flag Default Description
--output, -o query_doctor_report.html Output file path 
# Custom output path
python manage.py query_doctor_report --output=reports/turbo-dashboard.html

Schema Information

The report contains SQL query templates (without parameter values) that may reveal database schema information. Do not share the report publicly if your schema is confidential.

What the Dashboard Shows

Summary Cards

Five cards showing aggregate cache metrics:

Card Description
Cache Hits Total number of cache hits
Cache Misses Total number of cache misses
Hit Rate hits / (hits + misses) as a percentage
Cache Utilization current_size / max_size as a percentage
Evictions Number of LRU evictions

Charts

The dashboard includes Chart.js visualizations:

  • Cache Hits vs Misses — Doughnut chart showing the hit/miss ratio
  • Top Queries by Hit Count — Horizontal bar chart showing the top 10 most-hit cached queries
  • Prepared vs Non-Prepared Queries — Bar chart showing how many of the top queries use prepared statements (only shown when prepared statement data is available)

Top Optimized Queries Table

A sortable table of the top 20 cached queries, sorted by hit count:

Column Description
# Rank by hit count
SQL Preview First 200 characters of the compiled SQL template
Model The Django model label
Hit Count Number of cache hits
Prepared Whether the query uses prepared statements

Click any column header to sort.

When to Use

The dashboard is most useful after your application has been running for a while and the cache has warmed up. It provides insight into:

  • Whether the cache is being utilized effectively (high hit rate = good)
  • Which queries benefit most from caching
  • Whether poisoned entries indicate fingerprint collisions
  • Whether prepared statements are active (PostgreSQL + psycopg3)

Process-Local Data

The cache resets on server restart. The dashboard reflects the current process's cache state only. In multi-process deployments, each worker has its own cache.

Further Reading