A few years ago, AWS announced the release of their Scheduled Reserved Instances. These reserved instances are designed for workloads that recur on a daily, weekly, or monthly schedule, and are purchased for a one-year term. AWS says that Scheduled Reserved Instances provide a 5–10% savings over On-Demand instances used for this same purpose.
While we always appreciate ways to save on AWS, there are a few reasons that Scheduled Reserved Instances are unlikely to make a useful addition to your toolbox when compared to other cost-savings options available.
First of all, they have a decidedly limited use case, only for predictably scheduled operations that will go on for at least one year. Many companies would need to see a much higher savings rate than 10% with this year-long commitment when looking at AWS on demand vs reserved.
Secondly, they are inflexible. Once you set a schedule, you cannot change or override it, and the options to set schedules are limited to daily, weekly, or monthly recurrence on a set duration. Since one of the main benefits of cloud is the ultra-flexibility you get on short notice, this might be a deal-breaker by itself.
Additionally, as Beth Pariseau pointed in a TechTarget article, additional management overhead is required to manage every additional type of instance that a company leverages.
Note that Scheduled Reserved Instances are also limited by region — only available in US East (Northern Virginia), US West (Oregon), and Europe (Ireland) regions — and by instance type, currently supporting C3, C4, M4, and R3 instance types. This means that modern versions of those EC2 instance families, like M5, M5a, M5n, M6g, C5, or C5n, are not available for scheduling, so you’re using an older version of those EC2 instance types. While this might not matter much now, the list of usable instance types has not been updating along with the on-demand instance types, which means this problem will only get worse with time.
When compared to standard AWS reserved instance pricing, you’re not really getting the savings you’d expect from this kind of commitment. Reserved Instances typically save 30% for a convertible 1-year purchase, and can be about 60% for a standard 3-year purchase. This means that if you have an EC2 instance you need that would match the scheduled reserved instance but are using that same EC2 size for other workloads throughout the month, then the non-scheduled EC2 reserved instance pricing works much better with more savings.
For your recurring workloads, you can run instances only when you need them but maintain flexibility by using ParkMyCloud to schedule on/off times for On-Demand instances. By keeping workloads on just during business hours, you’ll save 65% using ParkMyCloud, with even more savings achievable if you need that instance even less throughout the month (and doesn’t even account for savings achieved through RightSizing). This beats any AWS RI pricing while maintaining flexibility for your organization. Keep this in mind while you are evaluating your reserved instance vs on demand decisions.
See the chart below for a full comparison of using Scheduled Reserved Instances vs. using ParkMyCloud.
This comparison shows that AWS Scheduled Reserved Instances are unlikely to be worth any effort or investigation by your cloud operations team. ParkMyCloud provides more benefits and much higher savings with more flexibility and less commitment. Even standard AWS EC2 Reserved Instance pricing and savings can give you more bang for your buck.
If you’ve got workloads and servers that don’t need to run very frequently throughout the month, but you need to ensure they can be spun up at a moment’s notice, then ParkMyCloud can help you save money and enable your users for maximum cloud efficiency. Give ParkMyCloud a try for yourself — start seeing savings today.
Originally published at www.parkmycloud.com on January 14, 2020.