AWS Cost Management
Savings Plans
Flexible pricing model offering up to 72% savings on compute in exchange for commitment
AWS Savings Plans are a flexible commitment-based pricing model that offer discounts of up to 72% compared to On-Demand rates in exchange for committing to a consistent amount of compute usage (measured in $/hour) for 1 or 3 years. Unlike Reserved Instances, Savings Plans automatically apply to any matching compute usage across services, instance families, and regions, making them significantly easier to manage. For cloud engineers, understanding Savings Plans is essential for cost optimization interviews and real-world FinOps work.
Savings Plans Types - Compute vs EC2 Instance vs SageMaker
AWS offers three types of Savings Plans with different levels of flexibility and corresponding discount levels:
| Type | Applies To | Flexibility | Max Discount | Best For |
|---|---|---|---|---|
| Compute Savings Plans | EC2, Fargate, Lambda | Any instance family, size, region, OS, tenancy | Up to 66% | Most organizations - maximum flexibility |
| EC2 Instance Savings Plans | EC2 only (specific family in specific region) | Any size and OS within a committed family/region | Up to 72% | Stable workloads in a specific region with known instance family |
| SageMaker Savings Plans | SageMaker ML instances | Any instance family, size, region, component | Up to 64% | ML-heavy workloads with consistent SageMaker usage |
The key trade-off is discount depth vs flexibility. EC2 Instance Savings Plans give the deepest discount but require you to commit to a specific instance family in a specific region. Compute Savings Plans are more flexible but offer a slightly lower discount.
Compute Savings Plans are generally the recommended choice for most teams because they handle instance type changes, region migrations, and workload shifts automatically. The discount difference vs EC2 Instance Savings Plans is small (66% vs 72%) and the operational savings from not managing per-family commitments is worth it.
How Savings Plans Apply to Your Bill
Savings Plans work by reducing your hourly compute bill as usage is matched against your commitment. The mechanics are important to understand for interviews:
When you purchase a Savings Plan at $10/hour, AWS looks at your compute usage each hour and applies the Savings Plans rate to usage up to $10/hour of commitment. Usage beyond your commitment falls back to On-Demand pricing.
| Scenario | What Happens |
|---|---|
| Usage < commitment | You pay the full commitment. Unused commitment is wasted - no refund. |
| Usage = commitment | All usage covered at discounted rate. Perfect utilization. |
| Usage > commitment | Commitment covers usage up to $10/hr at discounted rate. Excess is On-Demand. |
Priority order when multiple discount mechanisms exist:
| Priority | Mechanism | Notes |
|---|---|---|
| 1 | Reserved Instances (RIs) | Applied first to matching usage |
| 2 | EC2 Instance Savings Plans | Applied next to matching EC2 usage |
| 3 | Compute Savings Plans | Applied to remaining eligible usage |
| 4 | On-Demand | Charged for any remaining usage |
If you have both RIs and Savings Plans, the RI is always applied first. This means your Savings Plans utilization will look lower if RIs are covering usage the Savings Plan would otherwise cover. Track both RI coverage and Savings Plans coverage together.
Commitment Terms and Payment Options
Savings Plans come in 1-year and 3-year terms, each with three payment options. The payment option affects the effective hourly rate:
| Payment Option | 1-Year Discount (approx) | 3-Year Discount (approx) | Cash Flow Impact |
|---|---|---|---|
| All Upfront | Higher discount | Highest discount (up to 66%/72%) | Large upfront payment, zero ongoing charges |
| Partial Upfront | Medium discount | High discount | Upfront + monthly charges |
| No Upfront | Lower discount | Medium discount | Monthly charges only, no upfront commitment |
Break-even analysis for Compute Savings Plans (approximate):
| Option | On-Demand Rate | Savings Plans Rate | Monthly Savings on $10k/mo | Upfront Cost to Break Even |
|---|---|---|---|---|
| No Upfront 1yr | $10,000 | ~$7,400 | $2,600 | None - immediate savings |
| All Upfront 1yr | $10,000 | ~$6,800 | $3,200 | Pay ~$81,600 upfront for year |
| No Upfront 3yr | $10,000 | ~$6,200 | $3,800 | None - immediate savings |
| All Upfront 3yr | $10,000 | ~$5,400 | $4,600 | Pay ~$194,400 upfront for 3 years |
For most companies, No Upfront 1-year Savings Plans offer the best balance of savings and risk. They avoid locking up capital, allow annual reassessment of commitment levels, and still provide substantial discounts over On-Demand.
Savings Plans vs Reserved Instances
Both Savings Plans and Reserved Instances offer commit-to-save discounts, but they work differently. Understanding when to choose each is a common interview topic.
| Dimension | Savings Plans | Reserved Instances |
|---|---|---|
| Commitment unit | $/hour of compute spend | Specific instance type in specific region |
| Applies to | EC2, Fargate, Lambda (Compute SP) | EC2, RDS, Elasticache, Redshift, OpenSearch |
| Flexibility | Automatically applies across families/regions/services | Convertible RIs offer some flexibility; Standard RIs are rigid |
| Discount depth | Up to 66-72% for compute | Up to 72% for EC2; varies by service |
| RDS coverage | No - Savings Plans do not cover RDS | Yes - RDS Reserved Instances available |
| Management overhead | Low - commitment in $/hr, auto-applies | Higher - must manage per instance-type inventory |
| Marketplace | Cannot be sold on Reserved Instance Marketplace | Standard RIs can be sold on Marketplace |
When to choose Reserved Instances over Savings Plans:
| Scenario | Use RI Because |
|---|---|
| RDS databases | Savings Plans do not cover RDS - must use RDS Reserved Instances |
| Elasticache, Redshift | No Savings Plans equivalent - RIs are the only commitment option |
| Very stable EC2 workloads | Standard EC2 Instance Savings Plans match RI discount but Convertible RIs can be exchanged |
| Need to resell unused capacity | Standard RIs can be listed on the Reserved Instance Marketplace |
Purchasing Strategy and Coverage Targets
A common mistake is committing too aggressively and ending up with low utilization. The recommended approach is to start conservative and increase commitments over time.
Use the AWS Savings Plans purchase recommendation in Cost Explorer:
# Get Savings Plans purchase recommendations via CLI
aws ce get-savings-plans-purchase-recommendation \
--savings-plans-type COMPUTE_SP \
--term-in-years ONE_YEAR \
--payment-option NO_UPFRONT \
--lookback-period-in-days SIXTY_DAYS
# Check current Savings Plans utilization
aws ce get-savings-plans-utilization \
--time-period Start=2024-05-01,End=2024-06-01
# Check Savings Plans coverage
aws ce get-savings-plans-coverage \
--time-period Start=2024-05-01,End=2024-06-01 \
--granularity MONTHLYCoverage target guidelines:
| Coverage Target | Risk Profile | Reasoning |
|---|---|---|
| 70-80% coverage | Conservative | Leaves buffer for workload reduction without wasted commitment |
| 80-90% coverage | Balanced | Good savings while maintaining flexibility for 10-20% fluctuation |
| 90%+ coverage | Aggressive | Maximum savings but risk of underutilization if workloads shrink |
Savings Plans cannot be cancelled after purchase. If your EC2 usage drops significantly (e.g. you decommission a large service), you will continue paying the hourly commitment rate for the full term even if you have nothing to apply it to. Always purchase based on your minimum expected usage, not average usage.
Interview Focus Points
- 1What is the difference between Compute Savings Plans and EC2 Instance Savings Plans? When would you choose one over the other?
- 2Savings Plans vs Reserved Instances - which would you recommend for a company running EC2, RDS, and Fargate workloads?
- 3What happens to your Savings Plans commitment if you shut down a service and your hourly spend drops below your commitment amount?
- 4How does the Savings Plans coverage metric differ from the utilization metric? Which should you optimize for?
- 5A team has 70% Savings Plans utilization - is that good or bad? What should they do about it?
- 6Can you cancel a Savings Plan after purchase? What are your options if you over-committed?
- 7How does AWS apply discounts when you have both Reserved Instances and Savings Plans active simultaneously?
- 8Walk me through how you would decide the right commitment amount when purchasing Savings Plans for a new environment.