AWS | Cloud | FinOps
AWS Cost Optimization in 5 Perspectives – Service Rightsizing 
Michel Zitman 01 Mar, 2023
Share 
In our previous blog we introduced you to the first perspective of five. We talked about Service Rightsizing and what rightsizing is all about. After determining your optimal instance size, the next step is to look at the second perspective: AWS Purchasing Model Optimization. Up to 70% of cost savings are achievable by committing your capacity for 1 or 3 years on specific AWS services including EC2, RDS and Redshift. Consumption Model Optimization helps you choose the appropriate strategy for you.
Perspective 2: Purchase Model Optimization
To keep our theory close to the actual practice, we will once again dive into the AWS platform and its Reserved Instances.
Let’s look again at the now familiar m5.xlarge instance type in the Frankfurt region. In the previous example, we used the costs associated with on-demand EC2 computing. From a technical point of view, Reserved Instances offer the same compute performance with the same configurations. The biggest difference between traditional instances and reserved instances however is that the period for which you want to use these instances is now fixed for a set amount of time, and optionally accompanied with upfront payment. For example, if you’re willing to commit for one year with all upfront payment, you can reduce the compute costs by over 41%. And again, if you were to optimize ten instances of this size in the Frankfurt region in your workload, you would save in excess of 8.000 dollars a year. So even though AWS allows for a very flexible approach to selecting your instances, committing to and reserving your cloud instances upfront can lead to a pretty serious saving.
Saving plans
While we are on the topic of AWS Purchasing, we propose some other areas to look at in Perspective 2. For instance, saving plans. We touched on reservations, but AWS also has another model available as of November 2019: the Savings Plan. If you are a frequent user of on-demand workloads, look at the available committed savings plans and continue to reduce your instance costs. The savings plan also apply to services such as AWS Fargate and Lambda.
Spot Instances
As described by AWS, Spot Instances let you take advantage of unused EC2 capacity in the AWS cloud. Spot Instances are available at up to a 90% discount compared to on-demand prices. This type of instance is very useful for test workloads or workloads that can deal with a sudden termination by Amazon.
Purchase Subscription Type
For some services, like Amazon Workspaces for example, it’s wise to look at the monthly model versus the hourly model, as it is often the case that the monthly model is a much more affordable option.
Bring-Your-Own-Licensing models
If you spend more than millions a year on the AWS platform, consider going for an enterprise discount program agreement. Especially if you intend to sign up for other licenses from software vendors via the AWS Marketplace.
Value Added Resellers
We offer our Cost Management services as part of the resell services of AWS. This can also be very beneficial, as our best practices from years of experience can reduce your cloud spend continuously.
How to optimize your AWS costs even further
Do you have any further questions around Cost or AWS Purchasing optimization? Feel free to contact us. Ready to learn more about the five perspectives? Read up on Perspective 3: Service Characteristics Optimization in the next blog. Need this knowledge on the go, or do you want to share it with a colleagues? Download the full original whitepaper here: ‘AWS Cost Optimization in Five Perspectives’.
I am passionate about cloud computing, but specially about what it enables. So as a cloud consultant and Oblivion’s Cloud Financial Management Practice lead I wholeheartedly believe that introducing and developing CFM/FinOps helps organizations to efficiently use the cloud, and thus contributes to the positive impact that technology has on our lives. But good food, wine and books are just as important! Needless to say, family and friends even more.
