RDS Extended Support

There is a new line on your AWS bill. It is avoidable.

If you run Amazon RDS for MySQL 8.0, AWS moved you onto Extended Support on 1 August 2026 and started charging by the vCPU hour, on every instance, including staging, dev and the replicas nobody remembers. It is not a penalty and nothing is broken: AWS is charging to keep patching an engine version that left standard support. Upgrading is what makes it stop. Start with a free read-only check of your fleet.

The chargePER vCPU HOUR

Billed per instance.

Year 1 to 20.100 per vCPU-hr
Year 30.200 per vCPU-hr
MySQL 8.0Billing since 1 Aug 2026
Applies toProd, staging, dev
AlsoRead replicas
Stops whenYou upgrade
What The Line Item Is

Nobody chose this. It switched itself on.

RDS Extended Support is what AWS charges to keep shipping security patches for a major engine version after it leaves standard support. Enrolment is automatic, so there is no email you missed and no setting you forgot. The first most teams hear of it is the invoice.

01 · WHEN

MySQL 8.0 crossed the line on 31 July 2026

AWS lists 31 July 2026 as the end of standard support for RDS for MySQL 8.0, with Extended Support year one pricing starting 1 August 2026. So the first affected invoice is the August bill, which most teams read in early September. MySQL 5.7 has been in Extended Support since March 2024 and moved to the higher year three rate on 1 March 2026.

MySQL 8.0MySQL 5.7
02 · WHERE

Find it in Cost Explorer

Open Cost Explorer, set the range to this month, group by Usage Type, and filter for usage types containing ExtendedSupport. Whatever appears is what you are paying. It is worth doing before anything else, because the number is usually larger than the mental estimate and it is spread across more instances than expected.

Cost ExplorerUsage Type
03 · THE TRAP

It bills per instance

The charge is per vCPU per hour on every enrolled instance, not once per account. Production is the one people think of. Staging, dev, the analytics replica and the read replica behind a reporting tool all bill at the same rate, and those are usually the instances nobody has looked at in a year. Fleets pay several times what the production instance alone would suggest.

StagingDevReplicas
04 · POSTGRES

It is not just MySQL

PostgreSQL runs the same calendar. Versions 11, 12 and 13 are all in Extended Support today, with PostgreSQL 11 on the higher year three rate since 1 April 2026 and 13 billing since 1 March 2026. PostgreSQL 14 leaves standard support on 28 February 2027, so if you are on 14 this is a planning problem rather than a billing one, for now.

PostgreSQL 11, 12, 1314 next
The Math

What it actually costs you.

AWS publishes 0.100 dollars per vCPU-hour for years one and two, rising to 0.200 dollars from year three, using US East Ohio as the worked example. Rates depend on the engine version, the Region and how long the version has been out of standard support, so check the RDS pricing page for your own region.

InstancevCPUsYear 1 to 2, per monthYear 3 onward, per month
db.r6g.large2About 146 dollarsAbout 292 dollars
db.r6g.xlarge4About 292 dollarsAbout 584 dollars
db.r6g.2xlarge8About 584 dollarsAbout 1,168 dollars
db.r6g.4xlarge16About 1,168 dollarsAbout 2,336 dollars
Worked at the published US East Ohio rate over a 730 hour month, per instance. A single db.r6g.2xlarge on MySQL 8.0 costs roughly 584 dollars a month from August 2026, and roughly double that from 1 August 2028 when year three pricing begins. Multiply by however many staging, dev and replica instances share the same engine version. To be fair to AWS: this is not a fine, it is the cost of someone continuing to backport security fixes to an engine the upstream project has retired.
Worth Sitting With

Three years of premium, for the same date.

Here is the part that changes the decision. Extended Support for MySQL 8.0 runs until 31 July 2029. Standard support for MySQL 8.4, which you can upgrade to at no additional charge, also runs until 31 July 2029. Both roads end on the same day. One of them charges you by the vCPU hour for three years to get there, and the other does not. Paying for Extended Support buys time to plan an upgrade, which is a perfectly reasonable thing to buy. What it does not buy is extra runway.
The 5 Minute Self-Check

Check this yourself. You may not need us.

This is the whole check. If it comes back clean, you are not affected and there is nothing to buy.

STEP 1

List every RDS instance

Open the RDS console and go to Databases. Add the Engine version column if it is not shown, and do this in every region and every account, not just the one you work in daily. Forgotten instances in a second region are common and they bill exactly the same.

STEP 2

Look at the engine versions

Flag anything on MySQL 8.0 or 5.7, or PostgreSQL 11, 12 or 13. Those are in Extended Support now. Note anything on PostgreSQL 14 as a plan-ahead item, since its standard support ends 28 February 2027. Include replicas: they show as separate instances and bill separately.

STEP 3

Confirm in Cost Explorer

Group by Usage Type and filter for usage types containing ExtendedSupport, over the current month. That gives you the real number rather than an estimate, broken down by instance so you can see which environments are contributing.

STEP 4

Decide what it is worth

If the total is small and an upgrade is already scheduled, paying for a couple of months is a sensible trade. If it is significant, or you have no upgrade date in the calendar, that is when it is worth talking. If nothing appeared at either step, you are current and you can stop reading.

The Upgrade

Why teams put this off, and how we handle it.

Nobody postponed a major version upgrade out of laziness. They postponed it because it can break things, and because the person who would test it is busy. That is the actual work, and it is what we take off your plate.

Step 1

Pre-upgrade assessment

We check for deprecated syntax and removed features, compare parameter groups against the new version defaults, and review driver and ORM compatibility for your application. This is where surprises are supposed to surface, on a clone, not in production.

Step 2

Test on a restored clone

We restore a snapshot into a temporary instance, upgrade that, and exercise your application against it. It costs a little compute for a few days and it is the cheapest possible place to discover a broken query.

Step 3

Staging first, then production

Staging goes first and stays on the new version long enough to be believed. Production follows in an agreed window, once staging has been quiet. No environment is upgraded on the same day it is tested.

Step 4

Blue/green cutover with rollback

We use RDS blue/green deployments where the engine supports it: a synchronised copy runs on the new version, and the switch happens in a short controlled window with the old environment still standing behind you if something looks wrong.

The Offer

Free fleet check, then decide.

FREE

The fleet check

Read-only access, every RDS instance across your accounts and regions, engine versions listed, what Extended Support is costing you today, what it becomes at the year three rate, and a recommended upgrade order. No charge and no obligation. If you are already current, we say so and that is the end of it.

THEN

A fixed-scope upgrade

If you want the upgrades done, we scope them from the check and quote a fixed price with a written timeline. We share full pricing before any commitment. You can also take the report and do the work yourself, which is a completely reasonable outcome.

CONTEXT

Part of a wider bill review

Extended Support fees on aging engines are one of the findings a broader AWS cost audit surfaces. If the RDS line is the symptom rather than the whole story, the audit is the better starting point and this check is one section of it.

Common Questions

RDS Extended Support, answered.

Dates and rates here follow the AWS version calendars and the RDS pricing page. Check your own region, since rates vary.

What is RDS Extended Support?+
Amazon RDS Extended Support is a paid offering that lets you keep running a database on a major engine version after AWS stops supporting it as standard. You get security updates for critical and high CVEs, bug fixes for critical issues, and normal support cases, and you pay an hourly fee per vCPU on every affected instance. AWS enrols eligible databases automatically once the standard support date passes, so nothing breaks and nothing announces itself. It is available for up to three years past the end of standard support, after which AWS upgrades the major version for you.
How much does RDS Extended Support cost?+
AWS publishes 0.100 dollars per vCPU-hour for the first two years and 0.200 dollars per vCPU-hour from year three, using US East Ohio as the worked example. The exact rate depends on the engine version, the AWS Region and how many calendar years have passed since standard support ended, so check the RDS pricing page for your own region. The charge is per instance and per vCPU, which is why an 8 vCPU instance lands near 584 dollars a month at the year one rate, and why fleets with staging, dev and read replicas add up faster than people expect.
How do I know if I am being charged?+
Open Cost Explorer, set the date range to cover the current month, then group by Usage Type and filter for usage types containing ExtendedSupport. Anything that appears there is being billed. You can also check the RDS console directly: filter your instances by engine version and compare against the AWS version calendar. If nothing shows up in either place, you are not being charged and you do not need us.
Can I just turn Extended Support off?+
You can change the enrolment setting at any time through the EngineLifecycleSupport parameter, and AWS applies it immediately with no downtime. The catch matters: if you disable enrolment on an instance that is already past its standard support date, AWS automatically upgrades it to the next supported major version. That is an unplanned major version upgrade on production, which is the exact risk most teams were avoiding when they postponed upgrading. So switching it off is not a way to keep the old version for free. The real choices are to upgrade deliberately, or to keep paying while you plan one.
How risky is a major version upgrade?+
Real, but manageable, and the risk is the honest reason most teams postponed it. Major versions change defaults, deprecate syntax, adjust parameters and occasionally break drivers or ORMs, so an upgrade needs testing rather than a maintenance window and hope. We de-risk it by testing against a restored clone first, running staging ahead of production, and using RDS blue/green deployments for the cutover, which builds a synchronised copy on the new version and switches over in a short, controlled window with a rollback path.
How long does an upgrade take?+
It depends on how many instances you have and how much application testing the change needs, so we scope it before quoting rather than guessing. The pattern is consistent though: assessment first, then staging upgraded and exercised, then production during an agreed window once staging has been stable. The database cutover itself is usually the shortest part of the whole engagement. We put the timeline in writing with the quote.
What does the upgrade service cost?+
The fleet check is free. We look at every RDS instance across your accounts read-only, list engine versions, show what Extended Support is costing you now and what it becomes at the year three rate, and recommend an upgrade order. If you want us to do the upgrades, that is a fixed-scope engagement quoted after the check, and we share full pricing before any commitment. If the check shows you are already current, we will tell you that and there is nothing to sell.
Start with the free check

Find out what you are paying. Then decide.

A read-only look at every RDS instance you run, what Extended Support is costing you now, what it doubles to, and the order we would upgrade them in. Free, and genuinely useful even if you never hire us. If the check comes back clean, we will tell you that.

30 min Free Consultation →