D1 Slow Query Monitoring ⏱️

Swazz features an integrated slow query monitoring system for Cloudflare D1 databases. This helps developers and administrators detect latency regressions and optimize database indexes.

Design and Architecture

Database operations in the edge coordinator are automatically monitored. The getDB wrapper intercepts all database interactions using a JavaScript Proxy. It monitors:

• Prepared Statements: Timing executions of first(), run(), all(), and raw().
• Database Executions: Timing executions of exec().
• Batch Queries: Timing operations grouped under batch().

Whenever a database query exceeds the configured threshold, Swazz triggers three notification channels:

1. Structured Log: Logs a warning via console.warn with serialized JSON detailing the query, duration, threshold, and timestamp.
2. Cloudflare Analytics Engine: Writes a data point to ANALYTICS_ENGINE (if bound) for dashboard visualization.
3. KV Cache: Appends the query metadata to the admin:slow-queries KV array in the session cache (limited to the 100 most recent queries) with a 24-hour TTL.

Configuration

You can configure the threshold via the environment variables or wrangler configuration:

Variable	Description	Default
SLOW_QUERY_THRESHOLD_MS	The query latency threshold in milliseconds. Queries slower than this are flagged.	200

Admin inspection Endpoint

Administrators can retrieve the list of recent slow queries using the admin secret:

Fetch Recent Slow Queries

GET /api/admin/slow-queries

Headers:

• Authorization: Bearer <ADMIN_SECRET> or X-Admin-Secret: <ADMIN_SECRET>

Example Response:

[
  {
    "event": "slow_query",
    "query": "SELECT * FROM users WHERE delete_requested_at IS NOT NULL",
    "duration": 284,
    "threshold": 200,
    "timestamp": "2026-07-06T08:01:28.636Z"
  }
]