AWS Savings Plan – Can You Reduce Your Cloud Costs Further?

Committing to an AWS Savings Plan is like walking a tightrope in your cost optimization efforts.

Every hour that your team uses fewer cloud resources than the contracted amount is an hour of lost value. And if you approach it carefully and end up undercommitting, you’ll wind up paying for the resources your team needs in the highest AWS price tier.

Luckily, automated rightsizing and cloud optimization offer a way out. 

Keep reading to find out how you could eliminate the effort around AWS Savings Plans and reduce your cloud bill even more.

What is an AWS Savings Plan?

AWS Savings Plan is a pricing strategy that provides lower costs on regular on-demand instances in exchange for committing to one or three years of consumption. A Savings Plan necessitates a one-year commitment of an aggregate per dollar per hour commitment.

You choose your own hourly spending limit. Up to that commitment, all compute consumption is accessible at a discounted cost. When you surpass your allotted threshold, AWS charges you the standard on-demand fee.

For example, if you continue to run workloads after your Savings Plan has ended and you have not renewed it, the additional consumption will be charged at the usual on-demand pricing.

3 types of AWS Savings Plan

1. EC2 Instance Savings Plan

If you’re planning to use Amazon Elastic Compute Cloud (EC2) instances, the EC2 Instance Savings Plan offers some pretty good price reductions (the same as Standard Reserved Instances).

The benefit is that this plan can be changed in terms of size, operating system, and tenancy. You can pay for a dedicated tenant if you have unique licensing, security, and performance needs.

EC2 Instance Savings Plans also allow you to change the size and operating system of your instance, while Reserved Instances forbid any meddling of this kind.

However, EC2 Instance Savings Plans are only available inside a specific instance family in a given area. For example, you can switch from a machine running Windows to the same type and size, only running Linux. But you won’t be able to transition between instance families.

2. Amazon SageMaker Savings Plans

These instances include:

• SageMaker Studio Note,
• Data Wrangler,
• On-Demand Notebook,
• Batch Transform,
• Processing,
• Training,
• Real-time Inference in Amazon SageMaker.

SageMaker Savings Plans can save you up to 64% in the best-case scenario. But note that they’re relatively less popular than the other two savings plans covered in this section.

3. Compute Savings Plans

AWS also offers Compute Savings Plans, which provide approximately the same discount amount (66% versus 72%) but with fewer limitations. The AWS Compute Savings Plans include options for instance family, location, operating system, tenancy, and even the individual compute service.

For example, you can move your discount from an EC2 m2.xlarge instance running Windows in the US East region to an AWS Fargate instance in the US West.

Compute Savings Plans offer users a number of alternatives for how to use the compute time, so you can take on more tasks and apply your original savings to them as well. Another advantage of Compute Savings Plans is that they open the door to avoiding cloud waste even if you acquire more capacity than you need for a certain project.

The Catch-22 of AWS Savings Plan

It doesn’t matter if you pick Compute Savings Plans or regular EC2 instance Savings Plans. You need to understand the tricky nature of AWS Savings Plans, you need to know how they work.

You buy a Savings Plan based on hourly commitment. AWS uses the Cost Explorer to automatically calculate how that commitment will look like as a monthly charge on your AWS bill.

Here’s an example:

Your On-Demand monthly bill for one EC2 instance is $2,000. Let’s say that the Savings Plan combination of term and payment options gives you 30% savings. Then the recommended commitment will be c. $1.92 per hour – or $1,400 per month.

Problems begin when the hourly resource usage of the specific EC2 machine families you’re using starts to fluctuate. 

If you commit to more resources than you need, every hour your teams use less than the committed level is a waste of some of the value you paid for. If you under-commit to a Savings Plan and constantly go over it, you’ll end up paying for extra resources in the highest pricing tier, On-Demand. 

When using the AWS Savings Plan, you still get a discount on the cost of whatever EC2 instances your teams provisioned. If you over-provision an EC2 instance, you won’t be taking full advantage of the Savings Plan discount program.

And unlike Reserved Instances, you can’t resell the capacity purchased via AWS Savings Plans in the AWS Marketplace. That’s why it’s smart to buy Savings Plans incrementally. This helps to avoid over-committing to more resources than your company gets to use each month. 

I shared some more insights about these two AWS offers in this article: Do AWS Reserved Instances and Savings Plans really reduce costs?

To get an AWS Savings Plan, you need to forecast demand

AWS Savings Plans are by no means easier to manage than Reserved Instances. To reap the full benefits of them, you need the same skillset and toolkit for forecasting your AWS costs. 

Savings Plans are a great option for accounts where you’re able to predict a minimum amount of usage for the entire duration of the commitment period (one or three years). If you’re dealing with an application that has an unpredictable baseline, it’s a good practice to start by gathering usage data and then gradually committing to Savings Plans.

The above might sound straightforward, but imagine doing that at the scale of the entire enterprise. 

Large organizations may easily end up investing a lot of time and effort into the planning and procurement of Savings Plans: 

1. Multiple project or application teams need to provide an estimate of how many compute resources they will require in the upcoming period. 
2. DevOps and/or FinOps teams then need to review these plans and confirm that the planned projects actually require the requested resources. 
3. Finally, the finance team needs to go over the specific solutions like Savings Plans and make sure that the reserved capacity matches the requirements (and won’t result in over- or under-committing).

Much of the above requires going over separate spreadsheets, reviewing historical cloud usage, and carrying out complex forecasting of the expected compute capacity needs.

If you’re running on an AWS Savings Plan right now, you’ve probably done most or all of these tasks. 

Is there a way to save up even when you’ve already committed to a certain level of resource usage? There is, and it starts with rightsizing.

AWS Savings Plan won’t optimize your cloud bill on its own

Savings Plans can help to reduce your AWS bill, but you’re still the one in charge of infrastructure optimization.

That’s why rightsizing is such a critical initiative. And if you’re running a large cloud environment, you need a solution that automates rightsizing tasks for you.

Tracking down which resources are running, in which families, and which teams own them are all time-consuming tasks. Trying to make sense of all the 400 EC2 instances AWS offers (together with their pricing schemes) is no small feat. It might take you days or weeks to analyze your inventory and utilization to learn which instances can be downgraded. 

An AI-based cloud cost optimization solution like CAST AI can quickly identify underutilized pieces of your infrastructure and downgrade or terminate assets without you having to lift a finger.

Using Savings Plans with Amazon EKS? There’s a way to cut more of your bill

This won’t get you out of your contract, but it’s the #1 option to get more out of your Savings Plan if you’re running EKS clusters.

You can get a free Savings Report to see how much you could potentially save as well as actionable suggestions to do so.

And if you want to free your team to focus on other tasks than cost optimization, there’s also the automated version. The tool will keep on optimizing your setup for optimal cost and performance automatically and you will get more out of your savings plan.

---

FAQ