What the New AWS Savings Plans Mean for You

AWS just introduced its Savings Plans, a new instrument for customers to save on their cloud spend by committing to a level of spending over an extended period. Currently, this role is filled by Reserved Instances (RIs), which have saved companies massive amounts off their cloud bills for years.

But RIs also require a lot of careful strategy if you want to maximize their potential. With Savings Plans, AWS hopes to make things more simple and straightforward so cloud teams can spend less time optimizing costs and more time innovating.

We’ll explain more about how we support Savings Plans below, but first: the basics.

The Basics of Amazon Savings Plans

When you purchase a Savings Plan, you commit to a certain amount of hourly spend over a 1- or 3-year term. Resources that fit the parameters of that Savings Plans are charged at the discounted rate until you reach that commitment threshold. Any usage above that mark is charged at the On-Demand rate.

Savings Plans are available in two different flavors:

  • EC2 Savings Plans cover a single family of EC2 compute instance usage (e.g., m5) in any single region regardless of size or operating system. Pricing for EC2 Savings Plans maps to pricing for Standard Reserved Instances, meaning they offer the most savings.
  • Compute Savings Plans cover (and can float across) all families of EC2 compute and Fargate in all regions. Pricing for Compute Savings Plans maps to pricing for Convertible Reserved Instances, which means they result in slightly less savings than EC2 Savings Plans, but offer much more flexibility.

Like RIs, Savings Plans:

  • Have the choice of 1- or 3-year terms
  • Provide for all, partial, or no up-front payment models
  • Cover On-Demand Capacity Reservations

The key difference between Savings Plans and RIs is how you access discounted rates. With RIs, you purchase discounted usage for a specific resource, such as m5.xlarge in USEast with a Linux OS and Default Tenancy. The amount of discount is determined by the term, payment schedule, and RI type (Standard vs. Convertible). The discounts for Savings Plans are also based on term, payment schedule, and type (EC2 vs. Compute), but the discounted rate is much more flexible. Instead of only using the discount for a relatively narrow set of circumstances, Savings Plans applies the discount to a larger bucket of options, such as all m5 instances in USEast for an EC2 Savings Plan or all compute across EC2 and Fargate for a Compute Savings Plan.

The differences become most pronounced when comparing RIs and Savings Plans of similar savings rates. Both Standard RIs and EC2 Savings Plans boast up to 72% savings, but EC2 Savings Plans aren’t limited by OS or Tenancy. As long as the spend is under the spend commitment, EC2 Savings Plans can apply to any EC2 resource in the right family and region, regardless of OS, Tenancy, or instance size.

Convertible RIs and Compute Savings Plans also offer similar savings (up to 66%). Like EC2 Savings Plans, Compute Savings Plans aren’t limited by OS, Tenancy, or instance size. The big difference is that all compute in EC2 and Fargate, regardless of family or region, can be covered. You can now take advantage of the discounted rate no matter which EC2 or Fargate resource you use.

With Savings Plans, the key limit to savings is set by how much you commit to spending each hour. Think about your usage commitment as a labeled spend bucket, such as $10/hr of m5 usage in USEast. It doesn’t matter if you use an m5.large or an m5.8xlarge, because any m5 usage is included at the discounted rate as long as you have room in the bucket. An m5.8xlarge just takes up more room.


What Amazon Savings Plans mean for you

Existing inventories of RIs still have significant value and can be modified and exchanged. Instead of an either/or approach, this can be done in concert with purchasing new Savings Plans to create optimal coverage with the lowest investment.
RIs will be consumed first to cover On-Demand usage. Once they’re exhausted, Savings Plans will kick in (if applicable).

For customers/organizations new to RIs and committed spend:

  • The flexibility of Savings Plans could simplify the process of crafting new rate optimization strategies.
  • Coverable workloads still require careful planning depending on the term, payment model, and risk tolerance (e.g., minimum desired utilization and minimum expected savings).

For customers/organizations with existing RIs:

  • Reservations retain value and will be supported throughout their life.
  • Modifying and exchanging RIs while also purchasing Savings Plans requires planning to achieve optimal savings.

The best practices and methodology for that planning is contained in FinOps, the ideal cloud operating model for cloud financial management and optimization. The platform is designed and built to facilitate FinOps, enabling organizations to:

  • Be aware of and understand cloud infrastructure costs.
  • Develop a plan and goals for financial optimization.
  • Execute the plan and develop processes to continue at scale.

A key part of FinOps is using a Crawl, Walk, Run approach. Instead of trying to purchase all of your Savings Plans in one massive investment, you should get a complete view of your cloud spending, then experiment with smaller, strategic Savings Plans purchases. Once you’ve proven the value of your strategy, you can strategically buy more.