AWS EC2 SLA: Understanding Amazon's Uptime Guarantee

Q: AWS EC2 SLA: Understanding Amazon's Uptime Guarantee

## AWS EC2 SLA Overview Amazon Web Services provides different SLA tiers for EC2 depending on your deployment architecture. Understanding these tiers is essential for planning your infrastructure. ### Current EC2 SLA Tiers | Deployment Type | SLA | Monthly Downtime Allowed | |----------------|-----|-------------------------| | Multi-AZ (2+ instances) | 99.99% | ~4.4 minutes | | Single Instance | 99.5% | ~3.6 hours | ### What Counts as Downtime? AWS defines downtime as when all running insta

Q: When is this useful?

Reference the AWS EC2 SLA when designing production architectures on AWS, negotiating enterprise agreements, calculating composite SLAs across multiple AWS services, or planning disaster recovery strategies.

Detailed breakdown of AWS EC2's SLA commitments. Learn about single-instance vs multi-AZ guarantees, service credits, and how to calculate your actual EC2 uptime.

Cloud Provider SLAs

Detailed Explanation

AWS EC2 SLA Overview

Amazon Web Services provides different SLA tiers for EC2 depending on your deployment architecture. Understanding these tiers is essential for planning your infrastructure.

Current EC2 SLA Tiers

Deployment Type	SLA	Monthly Downtime Allowed
Multi-AZ (2+ instances)	99.99%	~4.4 minutes
Single Instance	99.5%	~3.6 hours

What Counts as Downtime?

AWS defines downtime as when all running instances in a region (for multi-AZ) or the running instance become unreachable. Importantly:

Scheduled maintenance is excluded from SLA calculations
Single-AZ failures don't trigger multi-AZ SLA violations if other AZs are operational
Instance reachability is measured from AWS's perspective, not your application's health

Service Credits

If AWS fails to meet the SLA:

Monthly Uptime %	Service Credit
< 99.99% (multi-AZ)	10% credit
< 99.0% (multi-AZ)	30% credit
< 95.0% (multi-AZ)	100% credit

Practical Implications

The gap between the SLA guarantee and actual uptime can be significant. AWS does not guarantee 99.99% for a single instance — only 99.5%. This means:

A single EC2 instance can be down for 3.6 hours per month within SLA
For production workloads, you must use multiple AZs to get the 99.99% guarantee
Auto Scaling Groups across AZs are the recommended pattern

Beyond the SLA

Your application's actual availability depends on more than just EC2. Consider:

ELB health checks and routing (ELB has its own 99.99% SLA)
EBS volume performance (EBS has a separate 99.999% durability SLA)
Dependent services (RDS, ElastiCache, S3) each with their own SLAs
Application bugs are never covered by cloud provider SLAs

Use Case