How do I compare managed database costs across providers?

The calculator models compute, storage, provisioned IOPS, Multi-AZ HA, read replicas, backup retention, and committed-use discounts for RDS, Aurora, Cloud SQL, and Azure Database — with real 2026 list prices side by side. That is the full bill, not the headline instance price, which is where the surprises live.

What is the Multi-AZ doubling trap?

Enabling Multi-AZ HA roughly doubles the compute cost — you pay for a standby instance you do not directly use. It is the right choice for production availability, but many teams enable it by default on dev instances and pay double everywhere. The calculator shows the HA toggle's cost impact so the trade-off is explicit.

Why is backup retention an RDS-only surprise?

RDS bills backup storage beyond the free snapshot allowance, and long retention or frequent snapshots accumulate into a real line item that Aurora and Cloud SQL handle differently. The calculator includes backup retention in the monthly projection, so the "managed DB is expensive" comparison includes the charges people forget.

Managed Database Cost Calculator — RDS vs Aurora vs Cloud SQL vs Azure DB (2026)

Managed Database Cost Calculator

The managed-DB bill is not one number. Four providers, four different ways to bill the same PostgreSQL instance. AWS RDS doubles your compute the moment you check "Multi-AZ." Aurora throws in the standby for free but bills storage by the GB plus every million I/O. Cloud SQL and Azure DB sit somewhere in between — and all of them quietly charge for backup retention once you exceed one copy. This calculator models AWS RDS, Amazon Aurora, Google Cloud SQL, and Azure Database for MySQL/PostgreSQL side by side: compute, storage, provisioned IOPS, Multi-AZ HA, read replicas, backup retention, and committed-use discounts. No sales fluff — 2026 list prices.

📐 Database Configuration

Pick an engine, instance size, deployment topology, and storage profile. The engine models all four providers against the identical workload so the comparison is apples-to-apples. Prices are on-demand list rates in each provider's primary US region unless a region is selected.

Engine

Instance Size (vCPU · RAM)

Deployment Topology

Read Replicas

Commitment Term

Provisioned Storage (GB)

Storage Class

Provisioned IOPS (only if io1 selected)

Backup Retention (days)

Monthly Egress to Internet (GB, optional)

Scenario Preset

AWS Region (affects RDS & Aurora only)

📊 Monthly Managed Database Cost — 4 Providers Side by Side

💰 Total Monthly Cost by Provider

Provider	Compute/mo	Storage/mo	IOPS/mo	Backup/mo	Egress/mo	Total/mo	Annual

📋 Managed DB Pricing Reference (2026 list prices)

On-demand hourly compute for the instance tiers used. GCP = us-central1, Azure = East US, AWS = selected region (us-east-1 baseline). Storage is per-GB-month. IOPS is RDS io1 only. Multi-AZ behavior and backup billing differ per provider — see notes.

Tier	RDS $/hr	Aurora $/hr	Cloud SQL $/hr	Azure DB $/hr
2 vCPU · 8 GB	$0.171	$0.160	$0.186	$0.165
4 vCPU · 16 GB	$0.342	$0.320	$0.372	$0.330
8 vCPU · 32 GB	$0.684	$0.640	$0.744	$0.660
16 vCPU · 64 GB	$1.368	$1.280	$1.488	$1.320
32 vCPU · 128 GB	$2.736	$2.560	$2.976	$2.640
64 vCPU · 256 GB	$5.472	$5.120	$5.952	$5.280

Storage: RDS gp3 $0.138/GB-mo · Aurora $0.10/GB-mo (+$0.20 per 1M I/O) · Cloud SQL SSD $0.17/GB-mo · Azure Premium SSD $0.115/GB-mo. RDS io1 $0.20/GB-mo + $0.065 per provisioned IOPS-mo. Backup: RDS charges $0.095/GB-mo for retained backups beyond the first DB-size copy (modeled at 10% daily change); Aurora/Cloud SQL/Azure include backup free up to 100% of storage. Multi-AZ: RDS/GCP/Azure bill a standby instance (compute ×2); Aurora's standby is included. Source: provider pricing pages, July 2026. Verify against your region.

The Bill Nobody Reconciles Until the Finance Ticket Lands

Managed databases are the quietest line item in the cloud bill — until the same PostgreSQL instance that costs $380/month on one provider costs $610 on another for identical throughput. The spread isn't margin noise. It's four structurally different pricing models that each hide a different tax. This calculator surfaces all four at once so you can see exactly where the money diverges.

1. Compute — and the Multi-AZ Doubling Trap

On AWS RDS, checking "Multi-AZ" does not buy you redundancy for free. It spins up a synchronous standby in a second AZ and bills it as a full second instance. Your 4 vCPU instance at $0.342/hr becomes $0.684/hr the instant you tick the box — $500/month in compute alone, before storage. Google Cloud SQL regional HA and Azure zone-redundant HA do the same: you pay for the standby. Amazon Aurora is the exception — its standby shares the same distributed storage volume, so Multi-AZ adds zero compute. For a single HA instance, that single difference is often a $200–400/month swing between Aurora and RDS at the same size.

⚡ The Insight: If you're running one HA PostgreSQL instance and not using Aurora, you are paying a "standby tax" that Aurora customers don't. At 8 vCPU Multi-AZ, RDS compute is ~$998/mo vs Aurora ~$922/mo — and that gap widens as you add the storage and I/O differences below. Aurora's pitch isn't just "better availability" — it's "we don't double-bill your standby."

2. Storage — gp3 vs io1 vs Aurora's GB + I/O

General Purpose SSD is cheap everywhere: RDS gp3 at $0.138/GB-month, Cloud SQL SSD at $0.17, Azure Premium SSD at $0.115. At 500 GB that's $69–85/month — negligible next to compute. The trap is provisioned IOPS on RDS io1: $0.20/GB-month plus $0.065 per provisioned IOPS-month. Provision 12,000 IOPS on a 200 GB volume and the IOPS line ($780/month) dwarfs the storage line ($40/month) by 19×. Aurora bills storage at $0.10/GB-month but then layers $0.20 per million I/O requests — invisible until you have a write-heavy workload. A million-row bulk insert can quietly cost more in I/O than the instance does in compute.

3. Backup Retention — The RDS-Only Surprise

Here's the asymmetry nobody warns you about: RDS is the only one of the four that bills retained backups. AWS gives you backup storage free up to 100% of your database size — i.e., exactly one copy. Push retention to 7 days and, assuming ~10% daily change, you're now paying for ~0.6 extra DB-sizes of backup at $0.095/GB-month. On a 500 GB database that's ~$28/month; on a 2 TB database at 30-day retention it's north of $500/month — a line item that simply does not exist on Aurora, Cloud SQL, or Azure DB, all of which include automated backups up to 100% of storage for free. If backup cost shows up as your third-largest component, that's an RDS tell.

4. Read Replicas — Storage Shared or Not

Read replicas are where the four diverge again. A Cloud SQL or Azure read replica is a whole second instance: you pay compute and storage again. An RDS read replica is the same. Aurora replicas share the cluster storage volume — you pay only for the replica's compute, never a second copy of the data. At 3 read replicas on a 1 TB database, that storage-sharing difference alone is ~$330/month (3 × 1 TB × $0.115) that Aurora doesn't charge. For read-heavy scale-out architectures, Aurora's storage model compounds in your favor.

5. Committed-Use Discounts — The Real Lever

All four providers reward commitment. RDS and Aurora 1-year No-Upfront reservations land around 35% off compute; 3-year around 60%. Cloud SQL committed-use discounts (1-yr ~45%, 3-yr ~64%) and Azure reserved instances (1-yr ~42%, 3-yr ~61%) are in the same ballpark. The catch: reservations apply to compute only — storage, IOPS, and backup are always on-demand. So a 3-year RDS commitment on an io1-heavy, backup-heavy workload saves less than it looks, because the discounted compute is a smaller fraction of the total bill. This calculator applies the discount to compute and leaves the rest at list, which is exactly how the providers bill it.

Who Should Care Most

Profile	Typical Monthly Bill	What This Calculator Reveals
Startup, single AZ, 1 instance	$120–300/mo	Compute dominates. RDS vs Cloud SQL vs Azure are within ~10% — pick on ecosystem, not price. Aurora's standby-free Multi-AZ only matters once you add HA.
Production, Multi-AZ, 1 replica	$600–1,400/mo	Aurora pulls ahead on HA (no standby tax) and replica storage sharing. RDS backup retention + io1 IOPS are the sneaky lines. Cloud SQL/Azure sit between.
Read-heavy SaaS, 2–3 replicas	$1,500–4,000/mo	Aurora's shared-storage replicas save $300–700/mo vs RDS/Cloud SQL/Azure, which each bill replica storage. The gap widens with storage size and replica count.
Write-heavy, io1 + long retention	$2,000–8,000/mo	RDS io1 IOPS + backup retention can rival compute. Aurora's per-I/O model may be cheaper or pricier depending on write volume — model it, don't guess.