PostgreSQL major-version upgrades for AWS. Fixed price, done while you sleep.
Still on RDS or Aurora PostgreSQL 11–14? Extended Support surcharges are already on your bill, or start on 1 March 2027. We plan, rehearse and cut over your upgrade for a fixed fee. Every step has a rollback plan, and nothing goes live without your go.
Your PostgreSQL version has a price tag now
When a major version leaves standard support, AWS keeps it running under Extended Support and bills you for every vCPU, every hour. That includes Multi-AZ standbys and read replicas. Years 1 and 2 cost US$0.10 per vCPU-hour; from year 3 it's US$0.20.
A small 2 vCPU Multi-AZ database costs about US$290 a month extra in years 1–2, and about US$580 from year 3.
| Version | What happens | Date |
|---|---|---|
| PostgreSQL 11 | Extended Support ends; AWS upgrades it for you | 31 Mar 2027 |
| PostgreSQL 12 | Surcharge doubles (year 3 pricing) | 1 Mar 2027 |
| PostgreSQL 13 | Already in Extended Support | since 1 Mar 2026 |
| PostgreSQL 14 | Community end of life, then Extended Support billing starts | 12 Nov 2026, then 1 Mar 2027 |
Dates are for RDS for PostgreSQL (AWS release calendar). Aurora PostgreSQL follows its own, similar calendar; the assessment confirms the dates for your clusters.
One fixed price. No hourly surprises.
Assessment
Anyone on PostgreSQL 11–14
- Read-only scan of your instances
- Written report of blockers: extensions, replication slots, parameter groups, authentication
- Your Extended Support cost, per instance
- Upgrade plan and a fixed quote
- Credited in full if you book the upgrade within 30 days
Single instance
One RDS instance or Aurora cluster under 500 GB
- Everything in the Assessment
- Rehearsal on a copy of production
- Cut-over in your quiet window
- Written rollback plan
- 7 days of follow-up support
Replicas + PostGIS
Primaries with read replicas, or PostGIS / pg_partman
- Everything in Single instance
- Read replica rebuilds
- Extension upgrade paths, e.g. PostGIS 2.5 → 3.x
Bigger, self-managed on EC2, or several instances? Ask for a fixed quote. AWS charges for rehearsal instances go on your own AWS bill, usually a few dollars.
Four steps, and you approve the one that matters
Assess
You give us read-only access. We find what will break before it breaks: extensions without an upgrade path, logical replication slots, custom parameter groups, old drivers and authentication.Rehearse
We upgrade a copy of your production database end to end, time it, and give you a test endpoint for your application.Cut over
In a window you choose, usually your night, we switch production over using Blue/Green or logical replication to keep downtime short. Nothing starts without your written “go”, and every step has a rollback.Verify
We check row counts, extensions, query plans and statistics after the switch, then stay on hand for 7 days.
22 years of PostgreSQL, working from New Zealand while you sleep
pgcloudupgrade is a small team of PostgreSQL and DevOps engineers based in New Zealand. Our lead engineer has worked with PostgreSQL for 22 years, running production major-version upgrades on Amazon RDS, Aurora, self-managed PostgreSQL on EC2, VMware and OpenShift. Every engagement uses the same tested runbook and checks, so nothing is improvised on the night.
We work New Zealand evenings. That's overnight across the US and early morning in the UK and Europe: your lowest-traffic hours, when a cut-over hurts least.
What's included, and what isn't
Included: the upgrade itself; a rehearsal; a rollback plan; post-upgrade checks; 7 days of follow-up; a written runbook you keep.
Not included: application code changes; performance tuning beyond the upgrade; on-call or 24x7 support. We schedule work in agreed windows. We don't take emergency or same-day calls, and that's part of why the price is fixed.
Questions
How much downtime should we expect?
With RDS Blue/Green, the switchover itself usually takes about a minute. Logical replication can get close to that too. The rehearsal measures it for your database, so you know before the night.
What access do you need?
Read-only access for the assessment. For the upgrade, a dedicated IAM role scoped to the instances in scope, which you can revoke the moment we're done.
What if something goes wrong?
Every cut-over has a written rollback plan, and you take a fresh snapshot beforehand. If a check fails, we roll back and reschedule, at no extra charge.
We use PostGIS, pg_partman or logical replication. Is that a problem?
It's the most common reason AWS upgrades fail, and exactly what the assessment looks for. The Replicas + PostGIS package covers it.
Can't we just click “upgrade” in the console?
Sometimes, yes, and if the assessment shows that's safe for you, we'll tell you. When it isn't, you find out mid-upgrade, in production.
How do we pay?
The assessment is paid upfront. Upgrades are 50% at booking and 50% after the cut-over. We invoice in US dollars.
Is there a contract?
Yes: a short service agreement plus a one-page statement of work with the fixed scope and price.
Find out what your upgrade involves
Tell us your engine (RDS or Aurora), versions, and the number and size of instances. We reply within one New Zealand working day with next steps and a booking link for the assessment.
[email protected]