In our previous blog we talked about the Service Characteristics. We close this series with Perspective 5: AWS Service Pattern Analysis (Deviating patterns). This concerns actively slicing and dicing and doing thorough analysis on your cloud spend. In this case, it’s valuable to look at abnormal behaviour, such as unexpected peaks or deviations from the expected trend line. By actively doing this, you can identify costs and behaviours that you won’t find in checklists and frameworks. Take two practical examples:
1. Financial Anomaly.
A customer employs an application that is using an S3 bucket with a monthly spend of $300. So far nothing is out of the ordinary, right? However: that same bucket had generated a cost level of 3.5k dollars a month for API calls alone.
This is financial anomaly, especially compared to the size of the storage. We started diving into it together with the customer’s application team and found out that the code was not cost-efficient. It was pulling the bucket every minute. Since requests are a billing metric in S3, this drove the costs, but unnecessary since it wasn’t a requirement.
We prevented further overspending by optimizing the application code and changing the polling method.
2. Hacked behaviour.
Another example is an organization that had a website with a stable and predictable user base. However, the autoscaling group was blowing up at the backend. They asked for help, and by analysing the sources of the costs we ultimately found out that the website had actually been hacked.
The Amazon Machine Images (AMIs) used by the autoscaling group contained crypto miners. This was just something that was figured out by getting warned by the cost level. A lot of damage has been prevented thanks to actively monitoring where your costs are created and which channels are responsible for it. But of course: don’t rely on financial management tools alone for your security.
Conclusions & Takeaways
- Start with cost insights and awareness (tagging is key)
Cost optimization starts with accountability and insights. Start withtagging and make sure all AWS stakeholders in your organization are aware and have transparent insights in the costs that they are generating on the AWS platform.
- Focus on the quick wins (rightsize, zombie asset reduction, reservations)
When you start with optimizing, focus on the quick wins first. This means rightsizing, making reservations and the immediate reduction of zombie assets. These are all easy adjustments to immediately make a difference on your cloud bill#
- Approach Cloud Cost management from TCO perspective
Approach cloud cost management from a TCO perspective and don’t base it solely on reducing the AWS bill directly.
- Drive the full cloud cost management cycle (blog 1)
Make sure that you continuously drive the circle and maintain your improvements. It is a learning process for you and your team members. But, by making it a standard practice, it does become easier#
- Request help, there is always a business case for cost optimization
Last but not least: request help if you are struggling with these cost management applications. As you have read in the blogs, there is always a business case to be made regarding cost optimization. For every AWS service, there is a best practice and opportunities to prevent unnecessary cloud spending.
How to optimize your AWS costs even further
During this blog series we talked about different optimization possibilities. Do you have any further questions around Cost Optimization? Feel free to contact us. 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’.