Cloud Native / Cloud Services / DevOps / Sponsored / Contributed

Cloud Bill Risks of AWS Reserved Instances and Savings Plans

15 Jul 2021 12:00pm, by

Peter Santis
Peter is the chief revenue officer at CAST AI and has more than 20 years of experience in fueling growth at pre-IPO tech companies.

Estimating requirements for cloud resources is challenging. Pinterest once committed to paying $170 million for AWS services, only to later buy even more capacity at a much higher rate because of increased user activity — and spend $20 million more than forecast.

Reserving capacity in advance is tricky if you don’t know how much resources your business will need next month. What about one or three years into the future, though? This is the main problem with AWS Reserved Instances and Savings Plans.

Keep on reading to see why these products might be risky for your company and what to pick instead.

Recap: What Are AWS Reserved Instances?

Businesses go for Reserved Instances because they offer great discounts compared to the on-demand pricing. All they need to do is commit to a given cloud capacity for one or three years. That’s it.

In some cases, they can also get a guarantee from AWS that these resources will be available to them in a given hosting region. Generally, the larger the upfront payment, the larger the discount rate on cloud resources.

What Are the Potential Savings on Offer?

EC2 Reserved Instances offer discounts reaching even 75% compared to the on-demand rates. However, there’s something you should know before jumping on the RI bandwagon.

When purchasing capacity up front, you need to make sure your team uses it. Every hour your instance goes unused is lost to you in terms of financial benefits. To maximize your use of RIs, you need to have a clear understanding of what your team will need in advance.

Moreover, the discounts come at a price.

3 Risks of Reserving Cloud Resources

1. Requirements Changes

When committing to specific resources or levels of usage, you need to assume that requirements won’t change while the contract is running.

One year is a huge commitment in the cloud world. A lot of things might happen during that time that could affect your requirements. Making accurate estimates in this context is very difficult.

Remember that example I shared at the beginning? During one holiday season, Pinterest users spent so much time on the platform that the company saw its cloud bill soar over the initial estimates.

2. Vendor Lock-in

Signing a contract for Reserved Instances means you’ll become dependent on AWS. In other words, you’ll be running the risk of vendor lock-in. Using the services of other providers like Google Cloud or Microsoft Azure isn’t going to make sense — after all, you’ve already paid so much for AWS services up front, so you need to use them.

Switching will come with high costs. Why would you need to switch in the first place? Perhaps AWS, or any other provider, can’t offer your team a solution to a challenge waiting for you six months from the moment you sign the contract.

3. Lack of Flexibility

Things change fast in today’s world. The likelihood that your cloud requirements change in the future is very high.

When facing a new problem, your team might need more resources. If you already committed to a given capacity, you’d need to get even more. Or go somewhere else and end up with unused capacity that you’ve already paid for.

Most importantly, with Reserved Instances you lose the flexibility of scaling or the option to configure resources in multiregion/zone distribution.

Setting Up Reserved Instances Is a Challenge, Too

It’s difficult enough to keep up with the latest AWS services and select a good match from almost 400 EC2 instance types. Add to that the need to assess your application’s requirements and workload patterns.

Manual techniques are time-consuming and inaccurate, making the entire process risky.

Avoid using cloud bill analysis solutions to figure out how many Reserved Instances or Savings Plans you need. The information they offer may cause you to reserve capacity based on the instance types you’re currently using, even if their kind and size are incorrect.

If you reserve the wrong instances, you’ll inevitably find yourself over-provisioning your workloads and wasting lots of money in the long run.

AWS Savings Plans Aren’t Any Better

The lack of flexibility in Reserved Instances often gets solved with AWS Savings Plans. They’re just like RIs, only you’re committing to a set amount of compute your teams can use over one or three years.

In Savings Plans, you’re not committing to specific instance types or configurations. You might still fall victim to vendor lock-in, however. You pay for cloud resources that might not be relevant anymore in a year or three from now. Plus, if your estimates are wrong and you go over the capacity, you’ll have to pay extra using the on-demand rate.

Is There a Way Out?

Booking capacity up front for a reduced price seems attractive, but it’s a trap. Instead, consider cloud cost optimization strategies such as rightsizing, autoscaling and Spot Instances. You can stay flexible and trim your cloud bill by ensuring that your teams provision exactly what they need.

Before you commit to costly Reserved Instances, try this free Savings Report to discover how much money you could save while maintaining your flexibility.

Featured image via Pixabay

A newsletter digest of the week’s most important stories & analyses.