Savings Plans
Definition
Flexible pricing model offering discounts in exchange for consistent usage commitment, without being tied to specific instance types.
Use Cases
- Lyft: Reducing compute costs for variable, multi-service workloads while keeping flexibility to change instance families and adopt newer compute options — Purchased AWS Savings Plans to cover a baseline of steady compute spend, then continued to use Auto Scaling and instance diversification for the remaining variable demand (Lowered ongoing compute spend versus on-demand pricing while maintaining flexibility to shift workloads without being locked to a specific instance type)
- Airbnb: Optimizing cloud spend for large-scale production services with predictable baseline usage and seasonal spikes — Committed to a steady level of compute usage and used commitment-based discounts (such as Savings Plans/commitments) to cover the baseline, while using on-demand/elastic capacity for peak periods (Improved cost efficiency for always-on services while preserving the ability to scale during peak demand)
Provider Equivalents
- AWS: AWS Savings Plans
- Azure: Azure Savings Plan for Compute
- GCP: Committed use discounts (CUDs)
- OCI: OCI Commitment-based discounts (Universal Credits commitment / committed spend discounts)
Frequently Asked Questions
- What's the difference between Savings Plans and Reserved Instances?
- Savings Plans are a spend commitment (for example, $/hour) that can apply across eligible compute usage, so you can change instance families, sizes, and sometimes even services without losing the discount. Reserved Instances typically tie the discount to a specific instance configuration (and sometimes a specific region/tenancy model), which can be cheaper for very stable workloads but is less flexible if your needs change.
- When should I use Savings Plans?
- Use Savings Plans when you have a predictable baseline of compute usage (for example, always-on web/API tiers, steady batch processing, or consistent container workloads) but you still want flexibility to change instance types, move between compute options, or modernize over time. A common approach is to cover the minimum steady usage with a Savings Plan and leave the spiky or experimental portion on on-demand pricing.
- How much does Savings Plans cost?
- There is no separate fee; you commit to a minimum spend rate (such as $X per hour) for a term (commonly 1 or 3 years). Your bill is then discounted for eligible usage up to the committed amount, and any usage above the commitment is billed at regular on-demand rates. The effective savings depend on the term length, payment option (e.g., all upfront/partial/no upfront where applicable), and how consistently your eligible usage matches the commitment.
Category: cloud
Difficulty: intermediate
Related Terms
See Also