// Definition
Reserved Instances (RIs) are a cloud pricing model where you commit to using a specific instance type in a specific region for a 1 or 3-year term in exchange for a significant discount versus on-demand pricing. AWS, Azure (Reserved VM Instances), and GCP (Committed Use Discounts) all offer variants of this model. Discounts range from 30% (1-year, no upfront) to 66% (3-year, all upfront) on AWS EC2.
// Why It Matters
The core trade-off: you accept commitment risk in exchange for price certainty. If you commit to a 3-year Reserved Instance and the workload it covers is terminated in year one, you continue paying for unused capacity. If you commit correctly, you save 30–66% versus running the same workload on-demand.
Three RI types on AWS: Standard RIs (deepest discount, least flexible — can't change instance type); Convertible RIs (can exchange for different instance type/family, ~15% less discount); and Savings Plans (commit to a dollar amount per hour rather than a specific instance, applies across EC2/Lambda/Fargate automatically). For most organizations, Compute Savings Plans provide the best balance of flexibility and discount — see our detailed comparison of RIs vs Savings Plans.
Key mistake: purchasing RIs before rightsizing. Committing to a large instance you could right-size to a smaller one locks in waste at a discount. Always rightsize first, then commit. RI portfolio management — tracking utilization and coverage, purchasing at the right time, and converting between types — is a core FinOps practice at Walk and Run maturity stages.
// In Practice
Scenario: A company runs 50 m5.xlarge instances continuously for their production web tier. On-demand cost: $0.192/hr × 50 × 8,760 hours = $84,096/year. 1-year Standard RI (no upfront): $0.119/hr × 50 × 8,760 = $52,122/year. Savings: $31,974/year (38% reduction) with zero architectural changes. Purchasing process: 30 minutes in the AWS console.