.png)
Every RDS MySQL instance has Performance Insights enabled by default. Most engineers know it exists. Far fewer have looked at what they are actually paying for it — or realized that the product they have been using is being discontinued in June 2026.
This guide covers the complete cost picture: what the free tier includes, what the old paid tier cost, what the new replacement (CloudWatch Database Insights) charges, and the decision framework for whether your MySQL team needs to pay anything at all.
What Is RDS Performance Insights?
RDS Performance Insights is a database monitoring feature that visualizes database load in terms of Average Active Sessions (AAS) — the number of connections actively doing work in the database at any moment. The core view is a single chart showing database load over time, broken down by wait events (CPU, I/O, locks, network), SQL statements, users, and hosts.
Unlike CloudWatch metrics, which give you infrastructure-level signals (CPUUtilization, FreeableMemory, IOPS), Performance Insights operates at the database engine level. It shows you which specific SQL statement is consuming the most database time, which wait event is bottlenecking performance, and whether you have lock contention slowing down your application.
The product requires no configuration beyond enabling it. It runs lightweight data collection inside the database engine with negligible overhead on query performance. AWS states the overhead is less than 1% of additional CPU load, and independent testing supports this claim.
Supported engines for MySQL: RDS for MySQL 5.7 (still works, but note Extended Support charges if applicable), MySQL 8.0, MySQL 8.4. Performance Insights is not available on db.t3.micro instances, which have insufficient resources for the monitoring overhead.
Also read: RDS Encryption: Does Encrypting Your Database Add Cost?
What Does RDS Performance Insights Cost in 2026?
The answer depends on which tier you are on and whether you have migrated to the new CloudWatch Database Insights product. As of May 2026, the pricing landscape has three tiers.
Tier 1: Free (7-Day Retention) — The Default
Every RDS MySQL instance gets 7 days of Performance Insights data retention at no charge. This includes the full dashboard, DB load visualization, SQL statement breakdown, wait event analysis, and access to the Performance Insights API. The 7-day window rolls forward continuously — you always have the last 7 days available.
Cost: $0. No setup required, no opt-in needed. Performance Insights is enabled by default on most RDS MySQL instance types. The only exception: db.t3.micro and db.t3.small instances in older generations do not support Performance Insights.
Tier 2: Performance Insights Paid Retention (DEPRECATED June 30, 2026)
For teams needing more than 7 days of history, AWS previously offered paid retention of 1 month to 24 months. This is being deprecated on June 30, 2026.
Legacy paid tier pricing (US East, before deprecation): $1.50 per vCPU per month for the first month of retention, plus $0.0631 per vCPU per month for each additional month of retention. Pricing does not depend on the amount of data stored — only on vCPUs and retention period selected.
Worked examples (2 vCPUs = db.r8g.large, 4 vCPUs = db.r8g.xlarge, US East, monthly):
| Instance | vCPUs | Retention | PI Paid Cost/Month (legacy) | DBI Advanced Cost/Month (new) | Price Increase |
| db.r8g.large | 2 | 7 days | $0 (free tier) | $0 (Standard mode free) | None |
| db.r8g.large | 2 | 1 month | $3.00 | $18.25 (Advanced) | 6.1x more expensive |
| db.r8g.large | 2 | 15 months | $4.76 | $18.25 (Advanced) | 3.8x more expensive |
| db.r8g.xlarge | 4 | 7 days | $0 (free tier) | $0 (Standard mode free) | None |
| db.r8g.xlarge | 4 | 1 month | $6.00 | $36.50 (Advanced) | 6.1x more expensive |
| db.r8g.xlarge | 4 | 15 months | $9.52 | $36.50 (Advanced) | 3.8x more expensive |
| db.r8g.2xlarge | 8 | 15 months | $19.04 | $73.00 (Advanced) | 3.8x more expensive |
PI Paid legacy: $1.50/vCPU/month (first month) + $0.0631/vCPU/month per additional month. DBI Advanced: $0.0125/vCPU-hr x 730 hrs = $9.125/vCPU/month. All USD, US East (N. Virginia). Source: AWS re:Post verified examples and AWS pricing documentation, May 2026. PI paid tier ends June 30, 2026. Verify at aws.amazon.com/rds/performance-insights/pricing and aws.amazon.com/cloudwatch/pricing — rates change.
Tier 3: CloudWatch Database Insights Advanced Mode (The Replacement)
CloudWatch Database Insights (DBI) is the AWS replacement for the Performance Insights paid tier. It has two modes:
Standard mode: 7 days of retention. $0. Available to all RDS and Aurora instances. This is what you get automatically after June 30, 2026 if you take no action — equivalent to what you had on the PI free tier.
Advanced mode: 15 months of retention. $0.0125 per vCPU-hour for provisioned RDS instances ($9.125/vCPU/month). Adds fleet-wide monitoring, SQL execution plans (Oracle and SQL Server only), OS process analysis (requires Enhanced Monitoring), lock analysis (Aurora PostgreSQL only), and integration with CloudWatch Application Signals.
The pricing trap: DBI Advanced must be enabled at the cluster level, not the instance level. If you have a production Multi-AZ cluster with 1 primary and 2 read replicas, enabling DBI Advanced covers all 3 nodes. For a 3-node db.r8g.xlarge cluster (4 vCPUs each): 3 nodes x 4 vCPUs x $9.125/vCPU/month = $109.50/month. Compare to the old PI paid tier which you could enable per instance. Source: doit.com/blog transition guide, verified from AWS re:Post.
Also read: RDS Reserved Instances: 1-Year vs 3-Year Break-Even Across All Engines
Is 7 Days of Free Retention Enough for MySQL Performance Debugging?
This is the practical question that determines whether your team needs to pay anything at all. The answer for most MySQL teams is yes — 7 days is sufficient for the vast majority of active performance debugging scenarios.
Scenarios Where 7 Days Is Sufficient
Sudden performance degradation: your application slows down, you open Performance Insights, and you see a spike in DB load from a specific SQL statement that started running yesterday. Seven days of history is more than enough to identify the statement, see when the problem started, and correlate it with a deployment or data change. This is the most common use case and 7 days handles it completely.
Wait event analysis: a database is running hot and you want to see whether the bottleneck is CPU, I/O, or lock contention. Seven days of wait event history gives you enough context to distinguish a new problem from a pre-existing pattern.
Post-incident review: a P1 incident occurs on a Tuesday. Your investigation happens Wednesday and Thursday. Seven days covers the incident and the surrounding context completely.
Developer debugging: a developer notices a query is slower than expected during feature testing. They enable Performance Insights (or check the existing dashboard) and trace the slow query in real time or within the last few days. Seven days is always sufficient for this use case.
Scenarios Where You Genuinely Need Extended Retention
Seasonal regression tracking: a SaaS application experiences performance degradation each quarter when a large customer runs their end-of-quarter reporting batch. You need 30-90 days of history to compare this quarter’s performance to last quarter and identify whether the same pattern is recurring or getting worse.
Compliance and audit requirements: regulated industries (financial services, healthcare) may have contractual or regulatory requirements to retain database performance data for 12+ months. This is a legitimate use case for DBI Advanced.
Long-term query regression: a query that ran in 50ms six months ago now runs in 500ms. You want to pinpoint when the degradation started and correlate it with schema changes, data volume growth, or MySQL version upgrades. This requires more than 7 days of history.
Capacity planning: you want to show trend data to justify a database instance upgrade to leadership. A chart showing steadily increasing DB load over 3-6 months is more persuasive than a 7-day window. CloudWatch metrics can serve this purpose for infrastructure-level signals, but Performance Insights provides query-level granularity that CloudWatch does not.
RDS Performance Insights vs Enhanced Monitoring: What Is the Difference?
Many teams run both without clearly understanding what each one covers. They are genuinely complementary and address different layers of the database stack.
RDS Performance Insights
Operates at the database engine level. Shows database load in terms of SQL queries, wait events, users, and sessions. Answers the question: ‘What is the database doing and which SQL statement is responsible?’ Free for 7-day retention. Requires no agent. Data collection overhead: less than 1% additional CPU.
Enhanced Monitoring
Operates at the OS level. Shows CPU utilization per process, memory usage, file system activity, and network throughput at the operating system layer. Answers the question: ‘What is the EC2 host (the underlying compute) doing?’ Priced at $0.00 per month for 60-second granularity in us-east-1 — wait, actually Enhanced Monitoring is charged based on CloudWatch Logs ingestion at roughly $0.50/GB, which for most databases means cents per month. Requires an IAM role for the monitoring agent.
The key difference: Enhanced Monitoring can show you that MySQL is consuming 85% of OS CPU. Performance Insights can show you that MySQL is consuming 85% of OS CPU because a specific SELECT statement with a missing index is running 400 times per second. You need both for complete visibility. Enhanced Monitoring for the infrastructure view; Performance Insights for the query view.
For DBI Advanced: OS process analysis (previously available via Enhanced Monitoring integration) now explicitly requires both Enhanced Monitoring to be enabled AND DBI Advanced mode. It is not available in DBI Standard mode. This is one of the feature regressions in the transition from PI to DBI for teams who relied on the process analysis view.
Related: RDS MySQL reserved instances and instance sizing before committing
The Complete CloudWatch Database Insights Feature and Cost Comparison
Here is the complete side-by-side comparison of what each tier provides as of May 2026.
| Feature | PI Free (old) | DBI Standard (new) | DBI Advanced (new — paid) |
| DB load dashboard | Yes | Yes | Yes |
| Top SQL breakdown | Yes | Yes | Yes |
| Wait event analysis | Yes | Yes | Yes |
| Retention | 7 days | 7 days | 15 months |
| Fleet-wide monitoring | No | No | Yes |
| SQL execution plans | Yes (Oracle, SQL Server) | No | Yes (Oracle, SQL Server) |
| On-demand analysis | Yes (Aurora PG/MySQL, RDS PG) | No | Yes |
| OS process analysis | No | No | Yes (requires Enhanced Monitoring) |
| Lock analysis | No | No | Yes (Aurora PostgreSQL only) |
| CloudWatch integration | Partial | Full | Full |
| Billing unit | vCPU/month | Free | $0.0125/vCPU-hr ($9.125/vCPU/mo) |
| 4-vCPU instance/month | $0 (7 days) | $0 | $36.50 |
| Enabled per instance or cluster | Per instance | Per instance | Per cluster (all nodes) |
| Available after June 30, 2026? | No (deprecated) | Yes (default) | Yes (requires opt-in) |
Source: AWS Performance Insights pricing page, AWS re:Post migration guide, DoiT transition analysis, AWS official documentation. May 2026. Verify at aws.amazon.com/rds/performance-insights/pricing and aws.amazon.com/cloudwatch/pricing — rates change.
How to Calculate Your RDS Performance Insights Cost for 2026
Step 1: Determine Your Current Tier
In the AWS console, navigate to RDS and check the Performance Insights tab for any database instance. If it shows ‘Standard’ or ‘7 days’ retention and you are not paying anything beyond your instance cost, you are on the free tier. No action needed for the deprecation.
To find costs in Cost Explorer: filter by usage type containing ‘PI_LTR_FMR’ (first month retention) or ‘PI_LTR_AMR’ (additional month retention). Any non-zero charges here indicate you are on the paid PI tier and need to act before June 30, 2026.
Step 2: Calculate Your DBI Advanced Cost (If Needed)
Formula: (number of vCPUs per instance) x (number of instances in cluster) x $0.0125/hr x 730 hours = monthly DBI Advanced cost.
Example: 3-node MySQL cluster (primary + 2 read replicas), each db.r8g.xlarge (4 vCPUs): 4 vCPUs x 3 nodes x $0.0125/hr x 730 = $109.50/month for the cluster. Note: you cannot enable DBI Advanced on just the primary — all nodes are included. Verify at aws.amazon.com/cloudwatch/pricing — rates change.
Step 3: Decide Whether You Need Advanced Mode
You need DBI Advanced if: (1) you have compliance or audit requirements for more than 7 days of performance history; (2) you use execution plan analysis for Oracle or SQL Server; (3) you need on-demand analysis for Aurora MySQL or PostgreSQL; (4) you want fleet-wide performance monitoring across hundreds of databases; or (5) you are actively debugging long-term query regression trends.
You do not need DBI Advanced if: your team uses Performance Insights primarily for real-time query debugging and incident response, your databases are all within Standard support versions, and you have no retention requirements beyond 7 days.
Also read: RDS Reserved Instances: Engine-by-Engine Pricing and Commitment Guide
Is Enabling RDS Performance Insights Free? Does It Require a Restart?
Two of the most commonly searched questions about Performance Insights — and both have simple answers.
Does Enabling Performance Insights Cost Money?
The 7-day retention tier is free for all supported RDS MySQL instance types. Enabling Performance Insights for the first time on a db.r8g.xlarge adds zero cost to your bill as long as you keep the default 7-day retention. Only enabling paid retention (1+ months) incurs charges — and that paid tier ends June 30, 2026 in favor of DBI Advanced.
Does Enabling Performance Insights Require a Database Restart?
No. Enabling Performance Insights on an existing RDS MySQL database does not require a restart or any downtime. AWS applies the change immediately without modifying the running database engine configuration. This makes it safe to enable on production databases at any time without a maintenance window.
Enabling or disabling Performance Insights does not affect backup behavior, replication, or any other database operations. The only effect is whether the performance metrics collection is running.
Does Turning Off Performance Insights Delete Historical Data?
Yes. If you disable Performance Insights paid retention (or let it lapse on June 30, 2026 without migrating to DBI Advanced), all performance data older than 7 days is deleted. It is not archived elsewhere. If you need to preserve historical PI data beyond June 30, 2026, export it using the Performance Insights API before the deprecation date. The API endpoints for DescribeDimensionKeys and GetResourceMetrics will continue to function after June 30, but the data they return will be limited to the 7-day Standard mode window unless you are on DBI Advanced.
Frequently Asked Questions
1. What does RDS Performance Insights cost?
The free tier (7-day retention) costs nothing. The legacy paid tier ($1.50/vCPU/month for 1-month retention) is being deprecated June 30, 2026. The replacement, CloudWatch Database Insights Advanced mode, costs $0.0125/vCPU-hr (approximately $9.125/vCPU/month). For a db.r8g.xlarge (4 vCPUs): $36.50/month. DBI Standard mode (7-day retention) remains free. Verify at aws.amazon.com/cloudwatch/pricing — rates change.
2. Is RDS Performance Insights free?
Yes, for 7-day data retention. Performance Insights includes a 7-day rolling window of performance data at no charge for any supported RDS instance type. Beyond 7 days, you previously paid the Performance Insights paid tier (deprecated June 30, 2026). The replacement is CloudWatch Database Insights: Standard mode (7 days) is still free; Advanced mode (15-month retention, additional features) costs $0.0125/vCPU-hr.
3. What is the RDS Performance Insights deprecation?
AWS is deprecating the RDS Performance Insights paid tier (flexible retention 1-24 months) and the PI console dashboard on June 30, 2026. After that date, all databases default to CloudWatch Database Insights Standard mode (7-day free retention). Teams needing longer retention or advanced features must explicitly opt into CloudWatch Database Insights Advanced mode before the deadline or lose historical data. Source: aws.amazon.com/rds/performance-insights/pricing.
4. Is CloudWatch Database Insights more expensive than Performance Insights?
Yes, significantly. DBI Advanced costs approximately $9.125/vCPU/month ($0.0125/vCPU-hr). The old PI paid tier for 15-month retention cost approximately $4.76/month for a 2-vCPU instance. DBI Advanced for the same instance costs $18.25/month — 3.8x more. For 1-month retention, DBI Advanced ($18.25) vs old PI ($3.00) is 6.1x more expensive. Source: AWS re:Post verified examples, May 2026.
5. Does Performance Insights work with MySQL 5.7 in Extended Support?
Yes, Performance Insights and CloudWatch Database Insights work with MySQL 5.7 running in Extended Support. There is no compatibility issue between the monitoring product and the Extended Support status. However, running MySQL 5.7 incurs $0.200/vCPU-hr Extended Support charges (since March 2026) that dwarf the Performance Insights cost. Upgrading to MySQL 8.0 or 8.4 eliminates the Extended Support fee, which is the higher priority action.
6. Can I enable Performance Insights on specific instances within a cluster?
For the legacy PI paid tier: yes, you could enable it per instance. For the new CloudWatch Database Insights Advanced mode: no. DBI Advanced must be enabled at the cluster level and applies to all nodes — primary and all read replicas. You cannot selectively enable it on just the primary. This is an important cost consideration for multi-node clusters: a 3-node cluster pays 3x the per-instance Advanced rate.
7. Does enabling Performance Insights require a database restart?
No. Enabling or disabling Performance Insights on an existing RDS MySQL instance does not require a restart, a maintenance window, or any downtime. AWS applies the change immediately. This makes it safe to enable on production databases at any time.
8. What is ‘DB Load’ in RDS Performance Insights?
DB Load is the core metric in Performance Insights: the number of Average Active Sessions (AAS) in the database at any moment. A session is active when it has submitted a query and is waiting for a response. A db.r8g.xlarge with 4 vCPUs can typically sustain a DB Load of 4 AAS (one per vCPU) before CPU becomes the bottleneck. DB Load above the number of vCPUs indicates saturation.
