.png)
The standby instance in a Multi-AZ RDS deployment runs every hour of every day. It receives synchronous writes from the primary, stays perfectly up to date, and waits. It does not answer a single query from your application. Your application never connects to it directly. It has one job: to become the primary database if the original becomes unavailable. For that privilege, you pay double.
Whether double is worth it depends on a number that most teams have not calculated: the dollar cost of 60 seconds of unplanned database downtime. Once you know that number, the Multi-AZ decision becomes arithmetic.
What Does RDS MySQL Multi-AZ Actually Cost With a Reserved Instance?
Multi-AZ reserved instances are purchased at exactly 2x the Single-AZ rate for the same instance type. There is no discount for accepting the Multi-AZ constraint — you pay the full 2x premium even on a 3-year All Upfront commitment.
All rates below are verified from Vantage.sh (May 18-19, 2026, sourced from AWS API). Verify at aws.amazon.com/rds/mysql/pricing — rates change.
| Instance | Deployment | On-Demand/hr | 1-Yr No Upfront/hr | RI Savings | Monthly Cost (RI) |
| db.r8g.large | Single-AZ | $0.239 | $0.160 | 33% | $116.80 |
| db.r8g.large | Multi-AZ | $0.478 | $0.320 | 33% | $233.60 |
| db.r8g.xlarge | Single-AZ | $0.478 | $0.320 | 33% | $233.60 |
| db.r8g.xlarge | Multi-AZ | $0.956 | $0.640 | 33% | $467.20 |
| db.r8g.2xlarge | Single-AZ | $0.956 | $0.640 | 33% | $467.20 |
| db.r8g.2xlarge | Multi-AZ | $1.912 | $1.280 | 33% | $934.40 |
Source: Vantage.sh May 18-19, 2026, AWS API. MySQL engine, US East (N. Virginia). 1-year No Upfront RI. Multi-AZ = exactly 2x Single-AZ across all sizes and payment options. Monthly cost = rate x 730 hrs. 3-year RI: shows N/A for r8g on Vantage.sh — verify current 3-year availability at aws.amazon.com/rds/mysql/pricing. Rates change.
The RI discount percentage is identical on Single-AZ and Multi-AZ. Buying a Multi-AZ RI at 33% off on-demand still costs exactly 2x the Single-AZ RI at 33% off. Reserved instances reduce your hourly rate; they do not close the deployment type cost gap.
Also read: RDS Encryption: Does Encrypting Your Database Add Cost?
What Is the Multi-AZ HA Premium Per Month?
The Multi-AZ premium is the difference between the Multi-AZ and Single-AZ RI rates for the same instance. Here is what you are paying each month specifically for the standby instance:
db.r8g.large: $233.60 Multi-AZ RI minus $116.80 Single-AZ RI = $116.80/month HA premium. $1,401.60/year.
db.r8g.xlarge: $467.20 Multi-AZ RI minus $233.60 Single-AZ RI = $233.60/month HA premium. $2,803.20/year.
db.r8g.2xlarge: $934.40 Multi-AZ RI minus $467.20 Single-AZ RI = $467.20/month HA premium. $5,606.40/year.
These premiums are for compute only. Storage also doubles in a Multi-AZ deployment because the standby maintains its own EBS volume synchronized with the primary.
Storage premium calculation: 500 GB gp3 storage at $0.115/GB-month = $57.50/month for Single-AZ. Multi-AZ charges twice: $115.00/month. Storage premium: $57.50/month, $690/year for 500 GB. Verify at aws.amazon.com/rds/mysql/pricing — rates change.
Also read: RDS Reserved Instances: 1-Year vs 3-Year Break-Even Across All Engines
Is the Multi-AZ Premium Worth It? The ROI Break-Even Analysis
Framing the Multi-AZ decision as ‘double the cost’ misses the financial logic. The correct framing is: ‘What is the expected annual cost of the database being unavailable for 60 seconds during an unplanned failure, and how does that compare to the Multi-AZ annual premium?’
Step 1: Estimate Your Failure Probability
AWS does not publish exact RDS hardware failure rates, but independent analysis and AWS operational data suggest that a single RDS instance experiences a hardware-level event requiring failover approximately 0.3% to 0.5% of the time per year. That is one event per 200-333 instance-years. For a fleet of 10 production instances, you might reasonably expect one failover event per 20-33 years from hardware failure alone. However, AZ-level events (power, network, cooling) are correlated — they affect all instances in an AZ simultaneously and have occurred in every major AWS region at some point. Multi-AZ protects against AZ-level events; Single-AZ does not.
Step 2: Estimate Your Downtime Cost
For a Single-AZ instance that experiences a hardware failure: restore from the most recent automated snapshot. AWS takes automated backups daily for RDS. Restore time depends on database size: approximately 30-90 minutes for databases under 100 GB, longer for larger databases. Your application is unavailable for the duration of the restore.
For a Multi-AZ instance: automatic failover in approximately 60 seconds. The DNS endpoint is updated to the standby. Applications reconnect at the new primary. Zero data loss (synchronous replication). Zero manual intervention.
The downtime cost formula: hourly revenue or business impact x hours of downtime. For a $2M/year SaaS application generating approximately $228/hour: a 60-minute Single-AZ restore costs approximately $228 per incident. A Multi-AZ failover lasting 1 minute costs approximately $3.80 per incident.
Step 3: Compare to the Annual Premium
db.r8g.xlarge Multi-AZ annual premium (compute only): $2,803.20/year. Expected annual Single-AZ downtime cost (one incident per 5 years, 60-minute restore): $228 / 5 = $45.60/year. The Multi-AZ premium ($2,803) is 61x the expected annual downtime cost ($45.60).
The math changes if you also account for AZ-level events. AZ-level disruptions — which Single-AZ instances cannot survive — are rarer but longer (typically hours, not minutes). One AZ event per 5 years with a 4-hour outage and $228/hour cost: 4 x $228 / 5 = $182.40/year. Still well below the $2,803 annual premium.
The break-even hourly downtime cost for the db.r8g.xlarge Multi-AZ premium: $2,803/year / (one 60-min incident per year) = $2,803/hour needed to justify the premium on pure statistics.
The ROI math says Single-AZ is statistically cheaper for most individual instances. But the math also says that a database failure during your peak sales day, a payment processing outage during end-of-month billing, or a user-facing outage during a key enterprise demo can cost far more than $2,803 in a single hour. Multi-AZ is not insurance you buy because the expected value is positive — it is insurance you buy because the tail risk is unacceptable. For revenue-critical databases, that is almost always the right call. For internal tools and reporting databases, the math rarely supports Multi-AZ.
What Does Multi-AZ Not Protect Against?
Multi-AZ provides automatic failover for the specific failure modes it is designed for. Understanding what it does not protect against prevents false confidence in the HA setup.
Multi-AZ Does Protect Against:
EC2 host hardware failure affecting the primary instance. Availability Zone-level network or power events (the standby is in a different AZ). Planned OS patching and minor version upgrades (failover during maintenance minimizes downtime). Primary instance storage failure.
Multi-AZ Does NOT Protect Against:
Data corruption or accidental deletion: the standby receives synchronous writes and will mirror any data corruption or table drop immediately. The standby becomes a corrupted or empty database just as fast as the primary. Point-in-time recovery from automated backups is the tool for data corruption — not Multi-AZ.
Regional failures: if the entire AWS region has a widespread outage, both primary and standby are affected. Multi-region architectures or cross-region read replicas address regional failures; standard Multi-AZ does not.
Application bugs causing database overload: a runaway query or a schema migration that locks tables affects both the primary and (indirectly) the standby. Multi-AZ does not protect against application-layer incidents.
When Should You Definitely NOT Pay the Multi-AZ Premium?
The Single-AZ to Multi-AZ decision is often made incorrectly by default. Many organizations copy production IaC templates into development and staging environments, resulting in Multi-AZ running everywhere regardless of the business justification. Here is when Single-AZ is the correct choice and Multi-AZ is pure waste.
Development Databases
A developer running a MySQL RDS database for local testing and feature development has no business case for Multi-AZ. If the instance fails, the developer waits 30-60 minutes for a restore from snapshot and loses at most a few hours of development context. The $116.80 to $233.60 monthly premium for a db.r8g.large or xlarge Multi-AZ adds $1,400 to $2,800 per year per development database for zero measurable benefit.
Staging and QA Environments
Staging environments should be production-like in configuration but not necessarily in availability. A staging database failure means a delayed deployment or a test that needs to be re-run — not a revenue impact. Convert staging databases to Single-AZ and accept the occasional downtime. The savings fund production database optimization instead.
The Real-World Scale of Dev/Staging Waste
An engineering team with 8 developers, each with a dev RDS database, plus a shared staging and QA environment: 10 total non-production databases. If all 10 run db.r8g.large Multi-AZ unnecessarily: 10 x $116.80/month premium = $1,168/month, $14,016/year. That is $14,016/year paid to AWS for standby instances that serve zero purpose in non-production environments. Converting all 10 to Single-AZ takes 30 minutes of CLI work and saves that $14,016 immediately.
Reporting and Analytics Databases
A read replica used for reporting or analytics runs at Single-AZ by definition — read replicas are not Multi-AZ. If you need the underlying data source to be highly available, make the primary Multi-AZ. The replica itself does not need to be. Any database that can tolerate a restore-from-snapshot recovery process (typically 30-90 minutes for moderately-sized databases) and where that downtime has no revenue impact is a Single-AZ candidate.
What Is the Right Way to Purchase an RDS MySQL Multi-AZ Reserved Instance?
Multi-AZ and Single-AZ reserved instances are purchased separately and are not interchangeable. A Single-AZ RI does not cover a Multi-AZ instance, and a Multi-AZ RI does not cover a Single-AZ instance.
The Purchase Specification
When buying an RDS RI, you specify: engine (MySQL), instance family (db.r8g), deployment type (Multi-AZ or Single-AZ), region, term (1-year or 3-year), and payment option. The deployment type is a hard constraint — it cannot be changed after purchase.
What Happens if You Convert Deployment Type During a Reservation
If you purchase a Multi-AZ RI and then modify the instance to Single-AZ: the Multi-AZ RI continues billing but the reduced-rate coverage no longer matches your running instance. The RI will apply to any other Multi-AZ instance of the same type in the account. If no other matching Multi-AZ instance is running, the RI is effectively wasted for the remainder of the term.
If you purchase a Single-AZ RI and later enable Multi-AZ: the Single-AZ RI no longer covers the Multi-AZ instance. The instance runs at on-demand Multi-AZ rates while your Single-AZ RI continues billing. You need to purchase a separate Multi-AZ RI to get the discounted rate on the Multi-AZ configuration.
The Correct Purchase Sequence
Decide on the final deployment type (Single-AZ or Multi-AZ) for each instance before purchasing any RI. If you are evaluating whether to add Multi-AZ to a currently Single-AZ database, run the instance on-demand while making that architectural decision. Convert to Multi-AZ first, then purchase the Multi-AZ RI. Reversing the order wastes money.
Also read: RDS Reserved Instances: Engine-by-Engine Pricing and Commitment Guide
How Does the Multi-AZ Premium Interact With Extended Support?
MySQL 5.7 entered Year 3 Extended Support on March 1, 2026. The $0.200/vCPU-hr surcharge applies to both the primary and standby instances in a Multi-AZ deployment — AWS charges Extended Support on every running instance at that engine version.
For a db.r8g.xlarge MySQL 5.7 Multi-AZ (4 vCPUs each, 2 instances): Extended Support Year 3 = 2 instances x 4 vCPUs x $0.200/hr x 730 hrs = $1,168/month. The 1-year RI covering compute is $467.20/month. Your effective monthly bill: $467.20 compute RI + $1,168 Extended Support + $115 storage = $1,750.20/month. The Extended Support charge alone is 2.5x the compute RI.
The Extended Support premium doubles for Multi-AZ versus Single-AZ exactly as the compute does. Single-AZ: 1 x 4 x $0.200 x 730 = $584/month Extended Support. Multi-AZ: $1,168/month. The $584/month additional Extended Support for Multi-AZ is often overlooked when calculating the total Multi-AZ cost.
CRITICAL: If you are running MySQL 5.7 or PostgreSQL 11 on Multi-AZ in Year 3 Extended Support, upgrading the engine version is the highest-ROI action available — far more valuable than any RI optimization. Upgrading eliminates $1,168/month in Extended Support charges. The best RI optimization on the compute saves at most $350/month. Upgrade first, then reserve. Source: aws.amazon.com/rds/mysql/pricing, verified May 2026.
Multi-AZ DB Cluster vs Multi-AZ Instance: The Cost Difference
AWS offers two Multi-AZ configurations for RDS MySQL and PostgreSQL. Most FinOps guides do not distinguish between them — which leads to significant cost surprises.
Multi-AZ Instance (Standard)
One primary instance plus one non-readable standby in a second AZ. Billed at 2x the Single-AZ rate. Failover time: approximately 60 seconds. The default Multi-AZ option for all RDS MySQL engines.
Multi-AZ DB Cluster
One primary plus two readable standbys across three AZs. Billed at 3x the Single-AZ rate (three instances). Failover time: under 35 seconds. Available for MySQL 8.0.28+ and PostgreSQL 13.4+ only. Provides faster failover and readable standbys — the standbys can handle read traffic, unlike standard Multi-AZ standbys.
For db.r8g.xlarge MySQL 8.0: Multi-AZ DB Cluster cost = 3 x $0.478/hr on-demand = $1.434/hr = $1,046.82/month on-demand. With 3 separate 1-yr No Upfront RIs: 3 x $0.320/hr = $0.960/hr = $700.80/month. Compare to standard Multi-AZ instance: $0.956/hr on-demand = $698.08/month; with 1-yr RI: $0.640/hr = $467.20/month.
The Multi-AZ DB Cluster costs $233.60 more per month ($2,803/year) than the standard Multi-AZ instance for the same instance size. In return, you get two readable standbys (useful for read-heavy workloads that benefit from additional read capacity) and failover in under 35 seconds versus approximately 60 seconds. If you do not need readable standbys and 60-second failover is acceptable, the standard Multi-AZ instance at 2x Single-AZ cost is the correct choice.
Three Real-World Multi-AZ Cost Scenarios
Abstract principles matter less than concrete numbers. Here are three realistic production fleet scenarios showing the total annual Multi-AZ cost across compute, storage, and Extended Support where applicable.
Scenario 1: Early-Stage SaaS (2 Production Multi-AZ, 6 Dev/Staging Incorrectly Multi-AZ)
Production (correct Multi-AZ): 2x db.r8g.large Multi-AZ with 1-yr No Upfront RI. Annual compute: 2 x $0.320/hr x 8,760 = $5,606.40. Annual storage (200 GB each): 2 x 200 x $0.115 x 12 = $552. Annual production Multi-AZ cost: $6,158.40.
Non-production (unnecessary Multi-AZ): 6x db.r8g.large Multi-AZ on-demand (dev databases rarely get RIs). Annual compute: 6 x $0.478/hr x 8,760 = $25,123.68. What Single-AZ would cost: 6 x $0.239/hr x 8,760 = $12,561.84. Wasted Multi-AZ premium on dev: $12,561.84/year. Plus unnecessary storage: 6 x 100 GB Multi-AZ storage overhead = $0.115 x 600 x 12 = $828/year. Total unnecessary Non-production Multi-AZ spend: $13,389.84/year.
Total Multi-AZ fleet cost: $19,548. What it should cost with correct Single-AZ for dev: $6,158 (production) + $12,562 (dev on Single-AZ on-demand) = $18,720. But the real opportunity: put dev on Single-AZ AND buy 1-year RIs for stable dev databases. Dev RI: 6 x $0.160/hr x 8,760 = $8,409.60. Total optimized cost: $6,158 + $8,410 = $14,568. Annual savings from optimization: $19,548 – $14,568 = $4,980/year. Two actions: convert dev to Single-AZ, then reserve.
Scenario 2: Mid-Size Engineering Team (10 Production Multi-AZ, 5 Staging Multi-AZ)
Production (correct Multi-AZ): 10x db.r8g.xlarge Multi-AZ with 1-yr No Upfront RI. Annual compute: 10 x $0.640/hr x 8,760 = $56,064. Annual storage (500 GB each): 10 x 500 x $0.115 x 12 = $6,900. Annual production Multi-AZ cost: $62,964.
Staging (unnecessary Multi-AZ): 5x db.r8g.xlarge Multi-AZ on-demand. Annual compute: 5 x $0.956/hr x 8,760 = $41,862. What Single-AZ staging on-demand would cost: 5 x $0.478/hr x 8,760 = $20,931. Unnecessary Multi-AZ staging premium: $20,931/year. Plus storage: 5 x 500 GB Multi-AZ overhead = $0.115 x 2,500 x 12 = $3,450/year. Total unnecessary staging Multi-AZ: $24,381/year.
Optimization path: convert 5 staging databases to Single-AZ, purchase 1-year No Upfront RIs for staging at $0.320/hr. Staging optimized cost: 5 x $0.320 x 8,760 = $14,016 + storage $3,450 = $17,466. Saving on staging: $41,862 – $14,016 = $27,846/year on compute alone. Total fleet optimization: $27,846 + $3,450 storage saving = $31,296/year. From two CLI commands and five RI purchases.
Scenario 3: Enterprise MySQL 5.7 Multi-AZ in Year 3 Extended Support
10x db.r8g.xlarge MySQL 5.7 Multi-AZ with 1-yr No Upfront RI and Year 3 Extended Support active since March 2026.
Annual compute (RI): 10 x $0.640/hr x 8,760 = $56,064.
Annual Extended Support: 10 instances x 2 nodes x 4 vCPUs x $0.200/hr x 8,760 hrs = $140,160/year. This is 2.5x the RI compute cost.
Annual storage (1 TB each): 10 x 1,000 GB x $0.115 x 12 = $13,800.
Total annual cost: $56,064 + $140,160 + $13,800 = $210,024/year.
After upgrading to MySQL 8.4 (current supported): Extended Support eliminated = $140,160/year saved. The upgrade project pays for itself in under 2 months of eliminated Extended Support charges. This scenario illustrates why the correct optimization sequence matters: Extended Support first, then RI optimization, not the reverse.
How Usage.ai Optimizes RDS MySQL Multi-AZ Reserved Instance Purchasing
RDS MySQL fleets typically have three cost optimization opportunities that get addressed in the wrong order: teams purchase Multi-AZ RIs before checking whether the database genuinely needs Multi-AZ, optimize the RI term before checking whether the engine version is in Extended Support, and focus on compute costs before addressing the storage doubling.
Usage.ai Flex Reserved Instances analyzes your full MySQL fleet — Single-AZ and Multi-AZ instances separately — and surfaces these opportunities in the correct sequence. The platform first identifies Extended Support exposure (MySQL 5.7 instances in Year 3, flagged with exact monthly dollar impact). Then it identifies non-production Multi-AZ instances and surfaces the Single-AZ conversion opportunity with exact savings calculations. Only after both of those actions are complete does it purchase RIs on the remaining fleet, matching the correct RI type (Multi-AZ or Single-AZ) to each running instance.
The 24-hour analysis refresh ensures that if an instance is converted from Multi-AZ to Single-AZ (or vice versa), the RI recommendation updates within 24 hours rather than waiting for the next quarterly review cycle. If a Multi-AZ RI becomes underutilized because an instance was converted to Single-AZ, Usage.ai provides cashback on the unused portion. Fee: percentage of realized savings only — $0 if no savings are generated.
See how much you can save on RDS MySQL Multi-AZ and Single-AZ reservations
Frequently Asked Questions
1. How much does an RDS MySQL Multi-AZ reserved instance cost?
RDS MySQL Multi-AZ reserved instances cost exactly 2x the equivalent Single-AZ RI rate. Verified May 2026 from Vantage.sh (AWS API): db.r8g.large 1-yr No Upfront RI = $0.160/hr Single-AZ, $0.320/hr Multi-AZ. db.r8g.xlarge: $0.320/hr Single-AZ, $0.640/hr Multi-AZ. The Multi-AZ premium at 1-yr No Upfront is $116.80/month for r8g.large and $233.60/month for r8g.xlarge. Verify at aws.amazon.com/rds/mysql/pricing — rates change.
2. How many instances are in a Multi-AZ cluster for RDS?
Standard Multi-AZ (DB instance): 2 instances — one primary and one non-readable standby in a second AZ. Multi-AZ DB Cluster: 3 instances — one primary and two readable standbys across three AZs. Standard Multi-AZ costs 2x Single-AZ. Multi-AZ DB Cluster costs 3x Single-AZ because all three instances are billed separately. Standard Multi-AZ failover: approximately 60 seconds. Multi-AZ DB Cluster failover: under 35 seconds.
3. What is the discount on reserved RDS instances?
RDS MySQL reserved instances save approximately 29-34% on 1-year No Upfront terms and up to approximately 53-58% on 3-year All Upfront terms (verify at aws.amazon.com/rds/mysql/pricing — rates change). For db.r8g, the 1-year No Upfront RI saves exactly 33% (confirmed: $0.160 vs $0.239 on-demand for Single-AZ). The same percentage discount applies to Multi-AZ — 33% off the 2x on-demand rate.
4. Does the Multi-AZ reserved instance cover both primary and standby?
Yes. A single Multi-AZ RI covers both the primary and standby instances as a combined purchase. You do not purchase two separate RIs for Multi-AZ. The Multi-AZ RI rate is approximately 2x the Single-AZ rate, reflecting the fact that both instances are running 24/7. The RI discount applies to the combined Multi-AZ compute cost as a single billing unit.
5. Do reserved instance discounts apply to Extended Support charges?
No. Extended Support charges ($0.200/vCPU-hr for MySQL 5.7 Year 3 since March 1, 2026) are billed separately from compute and are not reduced by reserved instances. For a Multi-AZ instance with 4 vCPUs, the Extended Support surcharge is 2 instances x 4 vCPUs x $0.200/hr = $1,168/month. This charge applies on top of the RI rate and is not discounted. Source: aws.amazon.com/rds/mysql/pricing.
6. Can a Multi-AZ RI cover a Single-AZ instance?
No. Multi-AZ and Single-AZ reserved instances are separate products that only cover matching deployment types. A Multi-AZ RI will only apply to Multi-AZ instances in the same account, region, and instance family. If you convert a Multi-AZ instance to Single-AZ, the Multi-AZ RI no longer covers it. The RI continues billing and will cover any other matching Multi-AZ instances in the account.
7. How much does Multi-AZ storage cost compared to Single-AZ?
Multi-AZ doubles storage costs because both the primary and standby maintain separate EBS volumes. At $0.115/GB-month for gp3: 500 GB Single-AZ storage = $57.50/month; 500 GB Multi-AZ = $115.00/month. The $57.50 storage premium compounds: at 1 TB, it is $115/month; at 2 TB, $230/month. Provisioned IOPS (if used) also doubles. Verify at aws.amazon.com/rds/mysql/pricing — rates change.
8. Is Multi-AZ worth the premium for a development database?
No. Development databases have no business justification for Multi-AZ. If a dev database fails, the developer waits 30-90 minutes for a snapshot restore — a minor inconvenience, not a business impact. The Multi-AZ premium for a db.r8g.large dev database is $116.80/month, $1,401.60/year. Converting 10 development databases from Multi-AZ to Single-AZ saves $14,016/year with zero operational impact on development workflows.
