Reserved Instances vs Pay as You Go

For organisations looking to save on their monthly cloud spend, it might seem obvious to go for Reserved Instances. But is it always the best option?

The answer is – it depends.

What is Pay as You Go?

Pay as You Go, is typically the model we think of in cloud computing. You pay only for the computing power or storage you need, and you can scale up or scale down as required.

A good example is Disaster Recovery as a Service (DRaaS). DRaaS is highly changeable. Most of the time, very little resource is required in the cloud, then occasionally, you need to spin-up an entire IT infrastructure. It’s sporadic and unpredictable so it’s perfectly suited to a Pay as You Go model.

What is a Reserved Instance?

Reserved Instances are best suited to workloads with consistent usage. A Reserved Instance is a commitment to a provider, typically between 1-3 years, where you get a discounted rate for a specified level of usage.

This is a model used by all major cloud providers including Amazon Web Services (AWS), Microsoft Azure, and Google Cloud.

Providers offer significant discounts up to 75% and a range of available payment options. The more you pay upfront, and the longer you can commit, the larger the discount.

Knowing whether it is beneficial to have a Reserved Instance is simple. If you are running a Virtual Machine (VM) round the clock, every day and you have a guaranteed, predictable usage, then there are savings opportunities available with Reserved Instances.

So, which is right for you?

Both options have pros and cons. In the real world, most computing requirements aren’t the perfect use-cases described above. They are rarely completely predictable or entirely unpredictable. You will usually need to look carefully at your requirements to decide which option is best.

Reserved Instances

The biggest benefits of Reserved Instances are the substantial discounts.

But discounted rates are not the only benefit. When you purchase a Reserved Instance you are paying for guaranteed capacity. For example, if AWS is reaching capacity in a certain Availability Zone, the Reserved Instance holder will get priority over a user without a Reserved Instance.

One of the major downsides of a Reserved Instances is, once purchased, you can’t cancel it. You can only modify, exchange, or resell.

Pay as You Go

The main benefit of Pay as You Go over Reserved Instances is it is more flexible and requires less work up front. You only pay for what you use, and you can start and stop the service at any time.

Pay as You Go is also a great option when launching a service for the first time or when in the testing phase. It’s best to start with pay as you go until you have a thorough understanding of your workload and its usage requirements.

Further considerations for Reserved Instances

Right-sizing

To get the correct usage, it is best to carry out a ‘Right-Sizing’ process before committing to a Reserved Instance. Services such as AWS and Microsoft Azure will provide you with analytical data of your usage which you can then use to gauge the level of commitment for a particular workload. This will make sure you are not over or under-spec’d and are getting the best possible discount.

Savings plans

The AWS Savings Plan offers the same discounts as Standard Reserved Instances but are determined on spend rather than utilisation.

They can also be applied to different operating systems and tenancies compared with a Standard Reserved Instance, where you’d need separate Standard Reserved Instances for each operating system or tenancy.

Those with Convertible Reserved Instances will benefit most from using an AWS Savings Plan. Unlike with a Standard Reserved Instance, a Convertible Reserved Instance you can edit the aspects such as the size, region, tenancy but it must be done manually. With an AWS Savings Plan any savings are carried across automatically when changes are made to the Reserved Instance.

Conclusion

Unless you are testing a new service or just getting started in the cloud, Reserved Instances will likely be the preferred option for most. Workloads should be considered on a case-by-case basis, and there’s a bit of initial leg work that needs to be done on behalf of the purchaser but there are huge savings to be made with Reserved Instances.