3.3.2. Pricing Models

Pay per hour or second, no long-term commitment. Useful for unpredictable workloads or short-term testing.

On-Demand Instances are a flexible pricing option provided by AWS, allowing users to pay only for the computing power they utilize. There is no need to make a long-term commitment; you can start and stop instances as needed. This makes it ideal for those who have unpredictable workloads or require instances for short-term testing. You get charged based on actual usage, either hourly or by the second, depending on the service.

Think of On-Demand Instances like renting a car. If you only need the car for a few hours or a couple of days, you can rent it without any long-term lease agreements. You only pay for the specific time you used the car, just like you pay for the on-demand instances only when you are using them.

Commit for 1 or 3 years and get up to 75% discount. Good for steady state workloads.

Reserved Instances require users to commit to using an instance type for a period of 1 or 3 years. In return for this commitment, users can receive significant discounts—up to 75% compared to On-Demand Pricing. This pricing model is suitable for workloads that are predictable and are likely to be consistently used over the duration of the reservation.

Imagine you decide to subscribe to a gym for a year rather than paying for single sessions. By doing this, you often get a much better rate compared to pay-as-you-go memberships because you're promising to use the facility more frequently. Similarly, by reserving an Amazon EC2 instance for a year, you secure a lower rate because you commit to using it consistently.

Bid for spare AWS capacity at up to 90% discount. Ideal for flexible, interruptible workloads like big data and batch jobs.

Spot Instances provide an opportunity to use spare AWS computing capacity at considerably reduced prices—for up to 90% off the standard rate. However, this pricing model requires users to bid for these instances. If your application can handle interruptions (meaning it can pause or restart when the instance is taken back by AWS), this option is very cost-effective. Spot Instances are typically used for batch processing, data analysis jobs, or scenarios where you can tolerate the possibility of your instance being shut down.

Think of Spot Instances like paying for an airline ticket during off-peak hours. Sometimes, the airlines offer huge discounts to fill empty seats, but there’s a catch: you need to be flexible with your departure time. Similarly, Spot Instances let you take advantage of AWS’s excess capacity, but you need to be ready for the possibility that the instance might get terminated.

Flexible pricing model offering discounts in exchange for a commitment to consistent usage.

Savings Plans provide a flexible pricing scheme that offers significant savings for users willing to commit to a specified amount of usage (measured in $/hour) over a period of 1 or 3 years. The benefit of this plan is that it applies to multiple instance types and across regions, giving users the freedom to adapt their use without losing the discount associated with their commitment.

Consider a mobile phone plan where you commit to a certain amount of data usage each month. If you frequently go over that limit, your provider might give you a discount for agreeing to a contract. This way, you pay less but have the flexibility to use data however you need. Similarly, with Savings Plans, you commit to a consistent baseline of spending, which gives you discount benefits across a range of AWS services.

Running a t2.micro On-Demand instance might cost $0.0116 per hour, but with a 1-year Reserved Instance, it could drop to$0.007 per hour.

This example illustrates a specific cost comparison between the On-Demand pricing model and the Reserved Instance pricing model. If a user runs a t2.micro instance on an On-Demand basis, the cost is $0.0116 per hour. However, if they opt for a 1-year Reserved Instance instead, the hourly price drops to$0.007. This stark difference in pricing highlights the potential savings available when users commit to longer-term usage of specific instance types.

Think about buying coffee. If you buy a cup of coffee from a cafe every day, it might cost you about $4 per cup. However, if you buy a 10-visit coffee card, each cup may only cost you$3! The same concept applies here; when you commit to using a service long-term, you get a better rate.

--