What high availability actually costs
AWS Multi-AZ Cost Calculator
Quantify the Single-AZ vs Multi-AZ premium — RDS instance and storage doubling, one NAT Gateway per AZ, and cross-AZ replication — in one number. us-east-1 list pricing.
- No signup
- Instant results
- ~2 min
Rates reviewed
Your HA footprint
us-east-1list rates, pricing as of 2026-10-20. Source: AWS public pricing.
Going Multi-AZ roughly doubles the RDS instance and storage, adds a NAT Gateway per AZ, and introduces cross-AZ traffic. We model those three drivers so you can price resilience before you commit to it.
RDS databasesize it →
Multi-AZ runs a standby of the same size — the instance line roughly doubles.
Multi-AZ mirrors storage to the standby, so storage is billed roughly twice.
Availability Zones (Multi-AZ)
Single-AZ uses 1 NAT Gateway; Multi-AZ uses one per AZ for resilient egress.
Cross-AZ traffic added by HA
Standby replication and app-to-standby traffic crossing AZ boundaries at $0.02/GB round-trip.
Your Monthly HA Premium
Multi-AZ minus Single-AZ
Monthly HA Premium
live resultSingle-AZ vs Multi-AZ
The per-driver premium breakdown and a right-sizing-HA checklist unlock after email.
Get the Full HA Cost Breakdown
Per-driver premium detail and where teams over- or under-buy availability — emailed to you.
Breakdown unlocked!
Per-driver premium and guidance below.
HA Premium by Driver
Right-Sizing Your Availability
- Match HA to the tier: production databases warrant Multi-AZ; dev/test and stateless replicas usually don't.
- Consider Multi-AZ clusters: RDS Multi-AZ DB clusters add a readable standby, turning the HA spend into read capacity.
- NAT per AZ is optional for some tiers: if a workload tolerates single-AZ egress, you can trim NAT gateways.
- Aurora reframes the math: Aurora is natively multi-AZ without the instance-doubling model — worth comparing.
Want this applied to your real account?
High availability is a per-tier decision, not an account-wide default. In an engagement we map each workload's real RTO/RPO to the cheapest architecture that meets it — often Multi-AZ on the database, single-AZ elsewhere, and Aurora where the cluster model wins. Typical AWS cost reviews cut spend 20–40%.
Talk to an Architect →Who This Tool Is For
CTOs, platform engineers, and FinOps leads deciding where to spend on resilience. If "make it highly available" became an account-wide default, part of that spend is probably buying availability the workload doesn't need.
Why We Built This Tool
High availability isn't free, and the cost hides across three services: RDS doubles the instance and storage for a standby, resilient egress wants a NAT Gateway per AZ, and Multi-AZ introduces cross-AZ replication traffic. This calculator isolates the HA premium so you can decide tier by tier instead of paying for it everywhere by reflex.
What Problem It Solves
- HA by default. Multi-AZ applied everywhere quietly doubles database and networking lines — this shows the delta.
- Hidden cross-AZ. Replication and standby chatter cross AZ boundaries at $0.02/GB — modeled here.
- Tier mismatch. Dev, test, and stateless tiers rarely need Multi-AZ. Quantify what you'd save by right-sizing.
- Defensible numbers. Export-ready figures for the resilience-vs-cost conversation.
Compare engines with the Aurora vs RDS calculator, or roll HA into the whole-bill calculator.
Frequently Asked Questions
How much does RDS Multi-AZ cost versus Single-AZ?
RDS Multi-AZ runs a standby replica of the same instance size, so the instance charge roughly doubles, and storage is mirrored to the standby so it is billed roughly twice as well. There is no separate "Multi-AZ fee" — the cost is the duplicated instance and storage. A readable-standby Multi-AZ cluster costs more still but gives you a second usable node.
Where do these prices come from?
AWS us-east-1 list rates as of 2026-10-20 (blended): RDS ~$0.09/vCPU-hour and gp3 storage $0.115/GB-month; NAT Gateway $0.045/hour per gateway; cross-AZ data transfer $0.02/GB round-trip. Confirm against the AWS RDS and VPC pricing pages before budgeting.
Do I need a NAT Gateway in every AZ?
For fully resilient egress, yes — a NAT Gateway lives in one AZ, so a single gateway is a single point of failure and forces cross-AZ traffic. Production Multi-AZ designs put one per AZ. Lower tiers can often accept a single NAT to save the hourly fees, at the cost of egress resilience.
Is Multi-AZ always worth it?
For production databases holding state you can't lose, almost always — the premium buys automatic failover and durability. For dev, test, stateless services, and easily-rebuilt tiers, it is frequently over-buying. The right approach is per-tier: match each workload's RTO/RPO to the cheapest architecture that meets it.
How accurate is this?
It is a directional list-price model of the three biggest HA drivers. It excludes Reserved Instances, read replicas beyond the standby, backup storage, and app-tier instance spread. Use it to size the premium, then validate against your actual instance types.
