Services covered under Compute Savings Plan (AWS)
The savings from Compute Savings Plan (AWS) cover the three most popular computing services, EC2, Fargate, and AWS Lambda. However, it is worth noting that not all aspects are covered, and savings are only applicable to specific configurations.
Therefore, you should not completely rely on Savings Plans to reduce your cloud expenditure, but take comprehensive actions to cut down cloud spend.
- Amazon EC2: EC2 is by far the most holistically covered service under this plan. Compute Savings Plans apply to all EC2 families, instance sizes, revisions, OS, and tenancy. This also allows unparalleled flexibility, since the customer can jump between configurations as they see fit for the usage.
- Amazon Fargate: Savings Plan only covers the memory and compute aspects of Fargate, thus becoming the compute service with the least coverage, apart from the year-based commitment discount, which offers up to 50% savings compared to On-Demand pricing.
- Amazon Lambda: Savings Plan allows for a maximum of 17% discount on the duration and provisioned concurrency functions of Lambda. However, while the other two compute services maximize discounts based on yearly commitment, Lambda does not offer such discounts.
Difference Between AWS Compute Savings Plan and AWS Reserved Instances (RI) Plan
The fundamental difference between AWS Compute Savings Plan and RI Savings Plan is flexibility, with Compute Savings Plan being more flexible of the two. The key areas where the former offers more flexibility than the latter are the types of instances covered and the ability to make changes to configurations post-purchase.
Here are the key differences in some important parameters:
| Parameter | Standard Reserved Instances | Compute Savings Plan |
| Discount Driver | Instance Type | Dollar Spend Commitment |
| Services Covered | EC2 (all families, regions, etc.) | All EC2 + Fargate + Lambda |
| Instance Family Flexibility | No flexibility. Fixed at the time of purchase. | Flexible. Since commitment is driven by dollar spend, users can switch between configurations. |
| Instance Size Flexibility | Except for instances running Linux, all are fixed | All instances, irrespective of OS, are flexible in size |
| Commitment Term | 1 to 3 years, with the ability to sell unused RIs on the RI Marketplace | 1 to 3-year commitment period, with no provision to sell unused capacity |
| Operating System | Fixed at the time of purchase | Can switch post-purchase |
Difference Between AWS EC2 Savings Plans and Compute Savings Plan
Both AWS EC2 Savings Plan and Compute Savings Plan are subcategories of AWS Savings Plans, which themselves are an evolution of Amazon Reserved Instances.
The fundamental difference between the two lies in the scope of services covered and the subtle differences in discounted pricing, with discount percentages varying over a 3-year commitment period.
| Parameter | AWS EC2 Savings Plan | AWS Compute Savings Plan |
| USP | Simplest to work with, offers the highest discount for EC2 resources | More flexible than EC2 Savings Plan in terms of instance family, size, OS, and region, with slightly lower discounts |
| Average Discount over 1 Year | 38% | 29% |
| Average Discount over 3 Years | 58% | 51% |
| Instance Family | Fixed at the time of purchase | Flexible; can switch between instance families |
Pricing Model of AWS Compute Savings Plan
When you opt for a Compute Savings Plan (AWS), you commit to spending a specific dollar amount on AWS resources per hour — for example, $15/hour — over a standard 1-3 year term. You pay upfront, partially upfront, or don’t pay upfront at all for your discounted rate for your usage. Also, unlike Reserved Instances, you are not tied to a particular instance family. The maximum 66% discount applies to all compute instances you use, regardless of configuration.
However, if you exceed the dollar amount you committed, you are charged based on On-Demand pricing. Thus, it is important to gauge your usage before opting for a Savings Plan.
Benefits of Compute Savings Plan (AWS)
- Payment Option Flexibility: For your Compute Savings Plan, you can opt for either no upfront payment (paid monthly), partial upfront (generally half of the total amount), or full upfront at the time of purchase, giving organizations with budget constraints more working capital.
- Horizontal Flexibility: Since the savings plan is tied to hourly spend and not to a specific instance type, you get operational flexibility for instance types, Availability Zones, Regions, operating systems, etc.
- Automated Discount Application: Compute Savings Plan automatically applies discounts through AWS's allocation logic, saving you the hassle of manually identifying resources where discounts can be maximized.
Limitations of Compute Savings Plan (AWS)
- Lock-in: Since the commitment period is 1 or 3 years, and to maximize discounts, you need to opt for full upfront payment with a 3-year plan, it leaves organizations with little room for flexibility if there is a fluctuation in resource utilization. This also impacts the engineering team's bandwidth, as they first need to extensively study the usage before making a purchase.
- Inefficient resource provisioning: Many organizations, in order to maximize utilization, risk exceeding their committed hourly spend, often resulting in On-Demand pricing charges. Conversely, they may underutilize the discount plan. Thus, simplifying Savings Plan utilization becomes a challenge.
- Difficulty in Accurately Predicting Utilization: Compute Savings Plans are best suited for consistent usage and should only be opted for in such cases. However, deviations in workload can significantly impact the ROI of the plan.
Best Practices to Maximize Savings with AWS Compute Savings Plan
- Always run Savings Plans on an empty AWS account: Savings Plans should be purchased on accounts without pre-existing long-term commitments. For example, avoid applying a Compute Savings Plan on an account already locked into heavy Reserved Instance usage, to prevent overlap and wastage.
- Analyze workload before committing: Since you’ll likely be opting for a 3-year discount commitment with upfront payment (if you’re aiming for maximum discounts), it’s important to have detailed insights into your usage patterns. Use AWS Cost Explorer and Compute Optimizer for a few months before purchase to identify consistent workloads and right-size them.
- Use with Spot Instances: For workloads where it’s difficult to predict long-term usage, leverage Spot Instances. By combining Spot Instances with your Compute Savings Plan (AWS), you can maximize savings while maintaining flexibility.
- Visibility is key: Regularly review your cloud infrastructure and ensure maximum visibility with CloudKeeper Lens, which will help you make informed decisions to maximize ROI from your Savings Plan.
Frequently Asked Questions
Q1:Do Compute Savings Plan discounts apply to Amazon S3?
No, Compute Savings Plans only apply to EC2, Fargate, and Lambda usage. For S3 discounts, consider S3 Intelligent-Tiering or volume pricing tiers.
Q2:Are Compute Savings Plan discounts applicable to Spot Instances?
No, Savings Plans do not cover Spot Instances. For Spot discounts, use EC2 Instance Savings Plans or optimize bidding strategies.
Q3:Can I use AWS Credits to purchase a Compute Savings Plan?
Yes, AWS Credits can be used to buy Compute Savings Plans, but check your credit terms for restrictions.
Q4:Can I modify instance configurations after purchasing a Compute Savings Plan?
Yes, you can change instance families, sizes, or OS types without losing Savings Plan benefits—flexibility is a key advantage.
