Multi-AZ, storage, and a fee you may not know about
Why is AWS RDS so expensive?
RDS is a managed database, and most of the price difference against running your own is the management. The parts that surprise people are Multi-AZ doubling, storage that only grows, and a support fee that started appearing on bills for old engine versions.
An RDS bill is the instance hours, doubled if Multi-AZ is on, plus storage at about $0.115 per GB-month for gp2, plus backup storage above the free allowance at about $0.095 per GB-month, plus data transfer, plus Extended Support at $0.10 per vCPU-hour in year one if the engine version is past end of standard support. The Extended Support fee is the one that appears without anyone changing anything, and on an db.m5.xlarge it is about $292 a month.
What an RDS bill is made of
| Component | Rate | Comment |
|---|---|---|
| Instance hours | Per instance class | Doubled if Multi-AZ is enabled |
| gp2 storage | ~$0.115 per GB-month | Provisioned, not used. Never shrinks. |
| Backup storage | ~$0.095 per GB-month | Free up to 100% of provisioned storage |
| Snapshots | ~$0.095 per GB-month | Manual snapshots persist after the instance is deleted |
| Extended Support | $0.10 per vCPU-hour, year 1 | Doubles to $0.20 in year 3 |
| Cross-AZ data transfer | $0.01 per GB each way | Multi-AZ replication traffic is free; app traffic is not |
| Performance Insights | Free for 7 days retention | Long-term retention is chargeable |
Conservative list prices. Aurora prices differently again, with storage billed on consumption and I/O billed per request unless you are on Aurora I/O-Optimized.
Where the money usually is
- →Multi-AZ on non-production. Multi-AZ doubles the instance cost to buy an automated failover. Production deserves it. A staging database that can be restored from a snapshot in twenty minutes does not, and turning it off halves that line immediately.
- →Extended Support on an old engine. When a MySQL or PostgreSQL major version reaches end of standard support, AWS keeps it running and charges $0.10 per vCPU-hour for the privilege. On a db.m5.2xlarge with 8 vCPUs, that is $584 a month on top of the instance. Upgrading the engine removes the whole charge.
- →Storage provisioned once and never revisited. RDS storage cannot be shrunk. Once you allocate 1 TB you pay for 1 TB until you migrate to a smaller instance. Autoscaling storage makes this worse quietly, because it only ever grows.
- →No reserved instances on a database that runs forever. Databases are the most predictable workload in any account, which makes them the best possible reserved-instance candidate. A one-year no-upfront RDS reservation is roughly 30% to 40% off, and unlike compute, the baseline genuinely does not move.
- →gp2 where gp3 would do. The same 20% saving that applies to EBS applies to RDS storage, with the same absence of downtime.
Put numbers on it
Frequently asked questions
What is RDS Extended Support and why am I paying for it?
It is a fee AWS charges to keep running a database engine version past its end of standard support date, so that you get security patches on a version that would otherwise be unsupported. It is charged per vCPU-hour, starts at $0.10 and doubles in year three. It is automatic, so accounts running an older MySQL or PostgreSQL major version started paying it without any change on their side. Upgrading the engine version stops the charge.
Is Aurora cheaper than RDS?
It depends entirely on I/O. Aurora costs more per instance-hour and does not charge for provisioned storage, only for what you use, but standard Aurora charges per million I/O requests. A write-heavy workload can spend more on I/O than on compute. Aurora I/O-Optimized removes the per-request charge for a roughly 30% higher instance price, which is worth it above about 25% of your Aurora spend going on I/O.
Does stopping an RDS instance save money?
Partly, and only for a week. Stopping halts instance hours but keeps billing storage and backups, and AWS automatically starts a stopped RDS instance again after 7 days. To keep a database stopped you need a scheduled job restarting the stop every week, which most teams find is more fragile than deleting it with a final snapshot.
Can I reduce allocated storage?
Not in place. RDS storage only scales up. Reducing it means creating a new instance with less storage and migrating, typically with a dump and restore or with DMS. That is why storage autoscaling deserves a maximum you have actually thought about.
Keep reading
Prices checked against AWS list rates on 2 August 2026. AWS changes prices; treat every figure here as a close approximation rather than a quote.
£499 fixed. Free scan first. 20%+ found or it’s free.
The scan is read-only — a role you create and delete, no keys shared — and shows your estimated monthly saving before anyone talks about money.