Reserved Instances
Definition
Reserved Instances are cloud computing capacity purchased in advance at discounted rates, requiring a commitment period of 1-3 years for cost savings.
Use Cases
- Netflix: Reduce steady-state compute costs for always-on production services running on AWS. — Purchased Reserved Instances for baseline EC2 usage and combined them with Auto Scaling and On-Demand/Spot capacity for variable demand. (Lowered compute spend for predictable workloads while maintaining flexibility for traffic spikes.)
- Airbnb: Optimize costs for long-running backend services with predictable utilization on AWS. — Committed a portion of their stable EC2 footprint to Reserved Instances and used On-Demand capacity for the remainder to handle variability. (Improved cost efficiency for steady workloads and reduced exposure to On-Demand pricing for baseline capacity.)
- Dropbox: Control cloud costs during periods where parts of the workload remained on AWS and required consistent compute capacity. — Used Reserved Instances for predictable, continuously running EC2 workloads while managing growth and migration planning separately. (Achieved lower unit costs for committed compute compared to purely On-Demand purchasing.)
Provider Equivalents
- AWS: Amazon EC2 Reserved Instances
- Azure: Azure Reserved VM Instances
- GCP: Committed use discounts (CUDs)
- OCI: OCI Reserved Instances
Frequently Asked Questions
- What's the difference between Reserved Instances and Savings Plans?
- Reserved Instances (RIs) are a commitment tied to specific instance attributes (like instance family, region, and sometimes availability zone) and can provide discounts when you run matching instances. Savings Plans are a commitment to a dollar-per-hour spend; they generally apply more flexibly across instance types (and, depending on plan type, across services like AWS Fargate and Lambda). In practice: RIs are more specific; Savings Plans are usually more flexible.
- When should I use Reserved Instances?
- Use Reserved Instances when you have workloads that run consistently (for example, production databases, core APIs, or always-on application servers) and you expect that usage to continue for 1 or 3 years. A common approach is to reserve only the baseline (minimum) capacity you reliably use, and keep burst or seasonal capacity on On-Demand or Spot.
- How much does Reserved Instances cost?
- You pay a discounted rate compared to On-Demand in exchange for a 1-year or 3-year commitment. The effective price depends on the instance type, region (and sometimes availability zone), term length, and payment option (no upfront, partial upfront, or all upfront). Longer terms and more upfront payment typically increase the discount, but reduce flexibility.
Category: cloud
Difficulty: intermediate
Related Terms
See Also