In the sixth and the final article comparing the AWS Reserved Instances and AWS Savings Plan models, we will go over the most suitable use cases for both.
AWS Savings Plans are a great option for automatic coverage, unlike Reserved Instances for which AWS administrators and Finance Operations (FinOps) executives are tasked with monitoring usage. If you buy an RI for a particularly powerful system that is downgraded later, the RI will be left without use, unless you swap it with a brand-new RI or transform it into smaller-size units. You will shell out money for an unused resource and this is where AWS steps in with its tools for governance and automatic coverage.
With Savings Plans, this issue will not give you sleepless nights. If a size, region, class, or tenancy of an instance is modified, the Savings Plan is automatically fine-tuned to still deliver the desired savings.
So, if you can predict a stable workload without deviations in a prolonged interval, Standard Reserved Instances for EC2 will be a great option. The same goes for workloads unaffected by changes in infrastructure, seasonal peaks, or small one-time projects.
By analogy, Compute Savings Plans will work best for the intensive use of Lambda or Fargate, such as when microservices play a crucial role. Also, these plans are a great option if you can predict an immediate need to change the region, instance family, or operating system for your workload.
In case you are dealing with low-use workloads, scheduling can be a viable option as a cost-cutting measure. In most cases, these are hosted on non-production instances that are not used on weekends or during evening and night hours on business days.
In conclusion, there is no universally good solution when it comes to AWS Reserved Instances and Savings Plans. While advice and common sense are your best allies, much will depend on your current business needs and the ability to predict changes as you go.