6 places you can save on cloud computing costs

Paying for cloud is the cost of doing business, but your bill doesn't often need to be as high as you think. Nine in ten (91%) organizations struggle with cloud spend that is avoidable. In this article, I’ll cover six practical places you can look to keep these costs down.

Important: First, find out what you’re working with

If you don’t already have it, you need a proper tagging and labeling system in place first for your cloud resources. Without it, you’re flying blind, because:

• Spend is aggregated, so you don’t know how it breaks down by team, service, or project

• You don’t know how many resources you have

• You don’t know how much usage those resources are getting and when

So, before you even start thinking about altering the balance of cloud resources, you want to actually understand what you’re working with, and have some historical data as well.

The major cloud providers have tools you can use to examine your costs and figure out patterns. For example, AWS has AWS Trusted Advisor, Compute Optimizer, and Cost Explorer.

I’ve got the data. Now, how can I cut down on costs?

Great! Now you’ve got the figures, it’s time to determine what cost reductions are possible, and how much you’re actually going to save by implementing them. 

That last part is important. Why? Because I can tell you to follow the steps below, but your mileage is going to vary depending on how much you’re actually spending on infrastructure. That will determine how many people you want to assign to these cost reduction tasks and how much time you want to have them doing it (TL;DR: It’s all about the ROI.)

Right, let’s get to the good stuff: where to look to save cloud costs.

1. Identifying under-utilized systems to be right-sized

When you create a cloud resource, it might be overkill for your needs, or the need for said resource might drop over time. Right-sizing is when you go back and adjust the cloud resource to fit the actual usage it’s getting. This typically means resizing an instance type or size to your actual workload performance and capacity requirements.

2. Terminate resources that are no longer needed

Sometimes you just don’t need a cloud resource at all, so you can axe it. This might sound like something that doesn’t happen, but it does, again through fluctuating demand or someone just forgetting to terminate a test server somewhere. Searching for these and shutting them down is a good way to get rid of clearly avoidable cloud spend. 

3. Lock down who can create high cost resources, particularly GPU and AI services

You still want to enable the business, but you can use policy-as-code to transform financial rules into guardrails that stop users spinning up something expensive and overkill before it happens. 

4. Consider stopping non-production instances outside of work hours when not in use

If nobody’s using your non-prod instances during certain hours (such as development, staging, or QA instances), shut them down. That’s avoidable cloud spend right there. You can apply this to off-peak hours and weekends.

5. Implement autoscaling to match capacity to demand

One of the greatest things about the cloud is that you can automate a lot and scale things to meet actual use. You should leverage this elasticity and reduce capacity for applicable cloud resources when demand is low.

6. Use spot and reserved instances and savings plans

You can choose different types of cost models for your instances based on your needs. A good cloud engineer or architect will know which ones are available, and the pros and cons of using these (this stuff is covered extensively in your foundational to intermediate cloud certs, because the cloud providers really want you to know the payment options.) For example:

• Spot instances can cut your compute costs by up to 90%, and these are good for supplementing your on-demand or reserved instances. They’re good for flexible, fault-tolerant work like batch processing.

• Reserved instances mean you’re committing to having an instance for a certain amount of time in exchange for a discount, typically up to 72%. They’re good when you’ve got predictable, steady workloads.

Should you shut down, scale, get strict, or use savings plans?

These are really the questions you should be asking when looking at your cloud resources. Hopefully the above list proves helpful! Good luck on cutting those cloud costs.

Cloud spend is at an all-time high, but even leading organizations struggle to see real returns. Read Pluralsight's How to Close the Cloud ROI Gap to learn how to build the talent, strategy, and systems that unlock value from your business’s cloud investments.