The countdown is nearly over. On June 30, 2026, Amazon RDS will retire Performance Insights, the built-in query monitoring tool that millions of RDS and Aurora teams use to diagnose slow queries, visualize wait events, and trace execution plans.
If you haven't migrated to CloudWatch Database Insights yet, you're about to lose capabilities you rely on, and potentially pay significantly more for the ones you keep.
What Is Going Away
After June 30, Amazon RDS will no longer support:
- The Performance Insights console experience: the query-level dashboard and wait event visualization you use today
- Flexible retention: Performance Insights allowed 1-24 months of history; that flexibility ends
- The existing Performance Insights pricing model
AWS originally set this deprecation for November 30, 2025, then extended it to June 30, 2026. That extension is now expiring.
The Two-Tier Replacement: Standard vs. Advanced Mode
Performance Insights is replaced by CloudWatch Database Insights (DBI), in two tiers:
Standard mode (no additional cost):
- 7 days of data retention
- Basic monitoring metrics
- No execution plans
- No on-demand analysis
Advanced mode (additional fee, scales per vCPU):
- 15-month retention
- Execution plans and on-demand analysis preserved
- Costs estimated at $100+ per month per large instance, before any third-party tooling
If you take no action, AWS defaults your instances to Standard mode. For teams doing active query optimization or incident response, losing execution plans with only a 7-day window is a significant regression.
Who Is Most at Risk
Production RDS/Aurora PostgreSQL and MySQL shops. Losing execution plans removes your fastest path to diagnosing a slow-query regression.
Cost-sensitive teams on large instances. Advanced mode scales per vCPU. A 32-vCPU production cluster could easily exceed $200/month per instance just for the tier upgrade.
Teams with compliance or audit retention requirements. Standard mode's 7-day window is insufficient for any post-incident investigation that extends beyond a week.
Your Migration Checklist Before June 30
- Audit your current usage. Identify which instances have Performance Insights enabled and which teams depend on execution plans or historical analysis windows beyond 7 days.
- Decide Standard vs. Advanced by instance. Development and staging environments may be fine on Standard; production query-critical clusters almost certainly need Advanced.
- Run the migration via AWS Console or CLI. AWS has published step-by-step guidance on re:Post. The mechanics are straightforward; the tier decision is where most teams stall.
- Account for the cost delta in your budget. Advanced mode appears as a CloudWatch line item. Add per-vCPU estimates for every production instance opting in.
- Evaluate whether DBI Advanced is sufficient for your needs. If your team needs richer query attribution, index-level recommendations, or retention beyond 15 months, this migration is a natural moment to ask whether a dedicated database observability layer belongs in your stack.
The Deeper Problem: Reactive Monitoring Isn't Enough Anymore
The Performance Insights deprecation exposes something larger than a tooling swap. Most database monitoring tools, including Performance Insights and its CloudWatch replacement, are reactive. They show you what already went wrong.
As agentic AI workloads land on production databases, query patterns are becoming less predictable. AI-generated SQL, agent-driven queries, and variable-load autonomous workflows create performance volatility that point-in-time dashboards weren't designed to handle. A June 2026 arxiv paper on cost-aware agentic query execution frames the same problem from the research side: runtime cost optimization for agentic workloads requires a fundamentally different monitoring approach than traditional reactive dashboards.
The industry is moving toward autonomous database reliability: systems that detect emerging anomalous patterns, model the cost impact before incidents surface, and recommend remediations autonomously.
The June 30 deadline is a forcing function. If you're re-evaluating your database monitoring stack anyway, the right question isn't just "which tool replaces Performance Insights." It's whether you want like-for-like reactive dashboards or a platform that closes the loop autonomously.
Datapace's autonomous database reliability layer continuously models query cost and performance trends, surfaces autonomous recommendations for index tuning and query rewrites before incidents surface, and integrates with AI agent workloads to identify which autonomous queries are driving cost spikes, without vendor lock-in to CloudWatch's billing model. See how Datapace approaches autonomous database reliability. If agents are running queries against your production database, book a call and we will walk through it on your stack.
Sources: pganalyze - AWS Performance Insights deprecation · AWS re:Post migration guide · DoIt transition guide · arxiv cost-aware agentic query execution