Implementing FinOps on AWS

I will start this article with the following sentence: The cloud is expensive!

This is what I usually hear out there. However, many who repeat this phrase have not even used public clouds, and those who have had such an experience sometimes lack the necessary knowledge to enjoy the various managed services without excessive spending.

Well, in this article, I will talk about how to implement FinOps on AWS.

In a few words, FinOps means that you will implement a cost-conscious culture in your company, in which your teams will have the following visibility: service consumption through dashboards; resources identified through Tags; automation of auto-scaling routines for on-demand handling; budget limits implemented through control policies; and cyclical architecture reviews focused on modernization, resulting in cost reduction.

Example: Let’s say you have an application that runs at certain times of the day and is on a virtual machine consuming resources 24/7. One of the options to modernize it would be to migrate it to AWS Lambda, where the billing model is based on the number of requests and the execution duration of a code. Here, we are talking about Serverless computing, which will likely bring you cost reductions.

To be successful in implementing this strategy, several services and tools will be necessary, which together will give you full financial control plus the possibility of reducing costs. Below I will mention some of them:

Cost Allocation Tags

Tagging is the first and main practice you must implement in your AWS accounts. You can assign Tags to resources such as EC2 instances, S3 buckets, RDS databases, among others. They will help you identify who is responsible for these resources and categorize them by department, project, environment, or any other relevant attribute. However, you will need a combination of AWS services to ensure that Tags are being adopted from the moment resources are launched and even to audit them to ensure there are no “untagged” resources in your accounts.

Service Control Policies (SCPs)

With SCPs, you can create a set of access rules to define permissions on AWS resources. For example, you can restrict a group of users’ access to a certain type of EC2 instance and thereby avoid excessive spending.

AWS Organizations

To implement SCPs, you will need AWS Organizations, which will replicate your policies to all accounts in your organization. Additionally, this service allows for the consolidation of billing from multiple accounts, providing a clearer view of expenses.

AWS Budgets

With AWS Budgets, you can set spending limits and receive real-time alerts via email or other notification services when costs approach or exceed these limits, thus avoiding unexpected costs.

AWS Cost Anomaly Detection

Still speaking of unexpected costs, you can implement AWS Cost Anomaly Detection to help you identify unexpected or unusual spending patterns that might go unnoticed. The service relies on Machine Learning algorithms that analyze your historical cost base and indicate deviations.

AWS Config

With AWS Config, you can continuously monitor configuration changes in your resources (including auditing resources implemented without Tags). Furthermore, it will also serve to identify changes in resources made outside of your Infrastructure as Code (IaC) templates. A good way to launch cloud resources is using IaC, because in addition to documenting the environment, you will have a standard for creating resources in the cloud.

Amazon QuickSight

To visualize your data, you can create dashboards and interactive real-time reports through Amazon QuickSight. With this service, you can compare costs across different periods, accounts, services, or regions, helping you understand spending trends and patterns from different departments or business units, as well as identifying underutilized or overutilized resources.

Regarding underutilized or overutilized resources, AWS has services to help you deal with these situations:

• Spot Instance Advisor: Analyzes the usage history of this type of instance, helping you obtain it at the best possible price. For those unfamiliar, Spot Instances are EC2 instances with lower prices than conventional ones, but they are not available 100% of the time. However, this type of instance can be easily implemented in a “containerized” environment as it is fault-tolerant.
• Compute Optimizer: Provides recommendations to optimize your EC2, EBS, Auto Scaling, and Lambda resources. The service uses Machine Learning algorithms to analyze computational resource utilization patterns and provides recommendations for Rightsizing.
• AWS WorkSpaces Cost Optimizer: Provides automatic recommendations to ensure you pay only for the necessary resources, matching instance types with the actual needs of your users. AWS WorkSpaces is a fully managed desktop virtualization service for Windows and Linux.

Regarding modernization, I recommend seeking an AWS Partner, as only they have access to incentive programs that will grant credits to your account and help make your project viable.

Edge UOL is an AWS partner and has a team specialized in more than 200 AWS solutions. We can help you implement and manage these services so you can focus solely on your business.