Cloud Cost Optimization is something every organization must perform. Failure to be successful at this task will lead to significant impacts. I have never seen a client that hasn’t struggled with cost optimization. All three major cloud providers (Google, Microsoft and Amazon) have a version of the “Well Architected Framework” that includes cost as a specific item. It’s not enough to perform cost optimization once but regularly to determine if you’re efficiently using your resources and you don’t have cloud sprawl.  The best part of the cloud is flexibility. This means that you can do things rapidly and easily without deep support from IT. This is also the sharpest sword that organizations can cut themselves with. I often find myself saying “just because you can doesn’t mean you should”. Business units can rapidly launch and use very expensive services unless you have well defined and mature guardrails. With that in mind let’s look at some strategies for managing this issue.   

In the grand scheme cost reflects state. In order to get the state, you need to determine where you are exactly. The best way to accomplish this is to do a complete cost analysis across the entire cloud footprint. I like to think of cost optimization as a treasure hunt. This is the “X marks the spot” stage of the journey with the treasure map. Marking that spot allows us to plot a course to the treasure but it also will give us the path for future analysis. The flow is:   

  1. find the initial cost issues
  2. identify solutions
  3. document the path 

The next steps will then follow the same path to make sure you are either maintaining the initial changes or finding the drift and remediating it. It’s a regression of sorts. Setting up a cadence to perform these steps is critical to success. Cost never stops so we must be consistent with controlling it.   

Three of the biggest areas where we can find and reign in costs   

  1. Overprovisioned Compute 
  2. Automation Efficiency 
  3. Capacity Commitment 

These three items are in relation to each other as shown below 

This diagram shows us an average of 20% savings across the cloud spend. I’m sure you noticed that none of the other percentages are 20%. This is because these items aren’t equal to the entire spend. Generally, resources that will fall into this particular savings description include:   

    • AWS: Ec2, Fargate, Lambda, SageMaker, Analytics, Cache 
    • Azure: Database, Compute, Storage and Analytics 
    • GCP: Compute, Anthos, BigQuery, Database 

Overprovisioned resources can be reduced by 40-60% depending on the application architecture and sizing. A spending plan on those resources can save you 30% on average with some reaching as high as 72% on the specific resources. Lastly the automation efficiency can help you right size around the clock based on your specific business. These items in conjunction give you the most consistent spend at the most appropriate level. As mentioned above with the audit cadence you don’t just do it once it’s a constant process to truly maximize return. The maximum return isn’t just cost it’s also performance efficiency, operational excellence and, reliability. The goal is to maximize all the factors at the same time not any single facet. This doesn’t mean that we aren’t also going to help you identify and control cost for network, managed services, storage and all the aspects in your cloud. This article just wants to focus on common and tangible items that we see across the customer base.     

For a Cloud Cost Optimization Analysis, we look at your entire cloud spend and give you a recommendation for saving with functional actions to achieve those specific savings. Everyone is unique and we know that analysis that isn’t or cannot be acted on isn’t going to save you money and that’s where we work with you to make sure our advice is both specific and relevant to your current and future situation.