On-Demand Instance Pricing
The default pricing model when you use AWS EC2 to spin up virtual machine resources is the On-Demand Instance pricing model. This is a hugely popular option, and it is not just because it is the default. With this model, you are not committed to any baseline amount of usage. You can launch a virtual machine for seven minutes, and you are billed for just seven minutes of usage. AWS recommends this model for these users:
Users who do not need discounts for long-term commitments and want the flexibility of a pay-as-you-go model
Users who are working with applications that have short-term, spiky, or unpredictable workloads
Users who are working with brand-new applications that they need to test in AWS, possibly over a very short term
Reserved Instance Pricing
Reserved Instance pricing is no longer recommended as often as it once was because AWS has created the Saving Plan pricing model (covered later). Because the Reserved Instance pricing model is still offered (and still tested on certification exams), we need to cover it here.
When you use Reserved Instance pricing, you can save up to 72% off of the On-Demand Instance pricing model. AWS provides this discount because you are committing to a certain amount of computing horsepower over time.
You can choose between three payment options when using the Reserved Instance pricing model:
All Upfront: You pay for the entire reserved instance term with one up-front payment. This option provides you with the largest discount compared to On-Demand Instance pricing.
Partial Upfront: You make a low up-front payment and are then charged a discounted hourly rate for the instance for the duration of the reserved instance term.
No Upfront: You do not need to make any up-front payment, and you get a discounted hourly rate for the duration of the term.
Who would benefit from choosing the Reserved Instance pricing model? This is what AWS says:
Users who have stable and consistent workloads and can predict their computing needs over longer time frames
Users who want to save costs in exchange for longer-term commitments to AWS
Users who are under strict budgets and need predictability when it comes to AWS compute costs
Savings Plan Pricing
Just like Reserved Instance pricing, the AWS Saving Plans pricing option saves you money (up to 72%) over the default On-Demand Instance pricing model by having you commit to purchasing compute resources over a certain time frame. In fact, AWS recommends the use of Savings Plans over Reserved Instances. Savings Plan pricing offers you the flexibility to change your usage as your needs evolve. For example, with a compute savings plan, lower prices will automatically apply when you change from C4 to C6g instances, shift a workload from EU (Ireland) to EU (London), or move a workload from EC2 to Fargate or Lambda.
Currently, AWS offers three types of Saving Plan pricing models: the compute savings plan, the EC2 instance savings plan, and the AWS SageMaker savings plan.
AWS recommends the Saving Plan pricing option for the following users:
Users who have steady-state usage requirements and have committed resource needs from AWS
Users who want to take advantage of the very latest compute options from AWS while still saving money over the default On-Demand Instance pricing model
Note
An advantage to the Reserved Instance and Savings Plan models is that the compute resources are guaranteed to be available in the region you are working with. I have tried to spin up large on-demand EC2 resources in specific regions, only to have AWS inform me that the resources are not presently available.