Last November, Amazon Web Services launched AWS Savings Plans for EC2, Fargate, and Lambda. Unlike what the name may imply, Savings Plans doesn’t mean dollar accrual; rather, it’s a discount plan, where purchasing in bulk leads to cost savings. To participate, the program requires customers to commit to spending a specific dollar amount per hour over a one- or three-year term. In return, AWS will offer discounts that can be as much as 72% off of the original price. Herein lies the guessing game: If you commit to 30 hours at a discounted rate and your usage exceeds that allotment, any time exceeding the 30 hours will be charged at the regular on-demand rate.
To participate, end users choose between two plan types, Compute Savings Plan (more flexible, lower max potential savings at 66%) and the EC2 Savings Plan (less flexible, higher max potential savings at 72%). Other than the obvious difference in savings, the added flexibility refers to where EC2 instance usage can be applied. In Compute, EC2 instance usage can be applied to any instance family, size, region, operating system, or tenancy, while EC2 Savings requires preidentified commitment of which individual families and in which AWS region the discount will be applied to — similar to how reserved instances (RIs) have historically worked.
Bear in mind, reserved instances aren’t going away, and there are some key differences between these two options. Unlike RIs, the current AWS Savings Plans can’t be applied to Relational Database Service (RDS) instances, AWS Redshift, or ElastiCache services. If you leverage RIs for these services, you should continue using RIs. The main distinction between RIs and Savings Plans is the former commits to number of instances used (RIs), while the latter commits to a minimum dollar per hour spend (Savings Plans). The other big difference between the two types is that Savings Plans gives the customer the option of an upfront payment, between 50% to 99%, in order to burn up capital before the end of a fiscal period.
Although RIs and Savings Plans have distinct differences, there are critical similarities. AWS Savings Plans are in part meant to address the overhead complexity associated with RI. However, it’s unclear if that issue was solved. Like RIs, Savings Plans will move between linked accounts and gravitate toward the purchasing account. Furthermore, like RIs, unused discounts won’t accumulate.
What It Means
AWS Savings Plans may appear more enticing with added flexibility, but it doesn’t do away with complexity. You’ll now have to keep track of Savings Plan “waste” in order to minimize the amount of unused discounts. And no, this does not automatically just optimize costs for you. You’ll still need a robust cloud cost management strategy that supports visibility, optimizes usage, and enables governance across all your computing platforms. Now you’ve got to make sure your cost optimization tool supports and optimizes for the AWS Savings Plan, too. For more insight on cloud cost optimization, read these reports: