3.1.1.2. Understanding SLAs in AWS Context
3.1.1.2. Understanding SLAs in AWS Context
An SLA is a contract between AWS and you — but understanding how to compose SLAs across services is the real exam skill.
AWS SLA math: If Service A has 99.99% availability and Service B has 99.95%, a system requiring both services has a composite availability of 99.99% × 99.95% = 99.94%. Adding serial dependencies always reduces availability.
Parallel redundancy improves availability: If a single component has 99.9% availability (8.76 hours/year downtime), two parallel components have 1 - (1-0.999)² = 99.9999% (31 seconds/year). This is why Multi-AZ and Multi-Region matter.
Key AWS SLAs to know:
| Service | SLA | Notes |
|---|---|---|
| EC2 | 99.99% (Multi-AZ) | Per-region; single instance has no SLA |
| S3 Standard | 99.99% availability | 99.999999999% (11 9s) durability |
| RDS Multi-AZ | 99.95% | Single-AZ has lower SLA |
| Lambda | 99.95% | Automatic Multi-AZ |
| Route 53 | 100% | Only AWS service with 100% SLA |
SLA credits: AWS issues service credits (billing discounts) when SLAs are breached — but credits require you to file a claim. SLAs don't guarantee uptime; they define the financial penalty for downtime.
Exam Trap: "Five nines" (99.999%) allows only 5.26 minutes of downtime per year. Achieving this requires active-active multi-region with automated failover — not just Multi-AZ. If a question specifies a five-nines requirement, Multi-AZ alone is insufficient. The answer involves Route 53 health checks, multi-region deployment, and automated failover.
