Understanding Reserved Instances
A Reserved Instance (RI) is a commitment-based pricing model offered by cloud providers, most prominently AWS, that gives customers a significant discount on compute capacity in exchange for committing to use a specific instance configuration for a fixed 1-year or 3-year term. Azure offers an equivalent called Reserved VM Instances, and Google Cloud offers Committed Use Discounts. Unlike on-demand pricing, where you pay per second or hour with no commitment, RIs trade flexibility for savings that AWS states can reach up to roughly 72% for a 3-year, all-upfront commitment compared to on-demand rates.
How It Works
AWS offers two RI types: Standard RIs, which offer the deepest discount but lock the commitment to a specific instance family and region, and Convertible RIs, which allow exchanging the reservation for a different instance family during the term in exchange for a somewhat smaller discount. Payment options include All Upfront, Partial Upfront, and No Upfront, with the discount percentage increasing as more is paid upfront. Critically, an RI is a billing construct, not a specific physical or virtual server: it automatically applies its discount to any matching running instance usage in the account, or across linked accounts if Reserved Instance sharing is enabled within an AWS Organization.
Example Scenario
A company runs a steady-state fleet of 20 m5.xlarge EC2 instances around the clock supporting a core backend service with predictable load. Rather than paying on-demand rates continuously, the company purchases 20 three-year Standard RIs with an all-upfront payment, reducing its compute cost for that fleet by approximately 60% compared to on-demand pricing. The savings apply automatically to any matching m5.xlarge usage in that region without any additional configuration.
Reserved Instances vs. Savings Plans vs. Spot Instances
AWS Savings Plans offer a more flexible alternative, committing to a dollar amount of compute spend per hour rather than a specific instance family, automatically applying the discount across any instance type or even serverless compute like Fargate and Lambda. Spot Instances offer even deeper discounts, often 70 to 90% off on-demand, but the provider can reclaim that capacity with short notice, making them suitable only for fault-tolerant, interruption-tolerant workloads, unlike the guaranteed availability of a Reserved Instance.
Why Teams Use Reserved Instances
RIs are best suited for workloads with genuinely steady, predictable, baseline resource consumption, such as a database tier or core application servers that run continuously regardless of demand fluctuations. They form a core lever of any FinOps commitment-based savings strategy, typically layered on top of instances that have already been right-sized.
Trade-offs and Risks
The primary trade-off is reduced flexibility: a Standard RI is locked to a specific instance family and region for the full term, so if architecture changes, such as migrating from x86-based instances to ARM-based Graviton instances, the existing RI commitment no longer matches and its discount is wasted. Upfront or partial-upfront payment options tie up capital that could otherwise be deployed elsewhere. Over-committing is a real risk if workload footprint shrinks due to optimization, decommissioning, or a shift to serverless architectures, leaving paid-for capacity underutilized.
Best Practices
- Only commit the portion of compute usage that represents genuine steady-state baseline demand, leaving variable or bursty capacity on-demand or covered by Spot Instances.
- Use Convertible RIs or Savings Plans instead of Standard RIs when architecture is likely to evolve, trading a slightly smaller discount for meaningfully more flexibility.
- Monitor RI coverage and utilization dashboards regularly to catch under-utilized reservations before they expire wastefully.
- Right-size instances before purchasing reservations; committing to oversized instances just locks in waste at a discount.
- Ladder commitments across a mix of 1-year and 3-year terms to avoid over-committing the entire compute footprint to the longest, least flexible term.
Frequently Asked Questions
What is Reserved Instances?
A Reserved Instance is a cloud pricing model where a customer commits to a specific instance type in a specific region for a 1- or 3-year term in exchange for a discount of up to roughly 72 percent versus on-demand pricing.
How does Reserved Instances work?
Reserved Instances works by combining the components described in the sections above. The main page walks through the architecture, the typical use cases, and the trade-offs to weigh before adopting it.
Why does Reserved Instances matter?
Teams adopt Reserved Instances to ship faster, run more reliably, and reduce the cognitive load on engineers. The benefits, limits, and adjacent tools are covered in the body above.
When should you use Reserved Instances?
Use Reserved Instances when the problems it solves match what your team is hitting today. The page above outlines the signals that mean you should adopt it now, and the cases where a simpler approach is fine.
