Before writing this article, I had been using AWS for over a year. I joined a company that was hosting their main website and other portals on Amazon services.
We recently transitioned all operations and development internally and so naturally I decided to learn more about AWS by studying for their Cloud Practitioner exam. Before going any further let me tell you, the certification is worth it.
Here is what I learned.
Under EC2 pricing, there are 5 payment models.
1. On- Demand
This is the default pricing model that is used when you launch an EC2 instance. It has no up-front payment and no long-term commitment. You are charged by the hour or by the minute depending on the EC2 instance type you are using.
This model is best used for applications that do not need to run for very long. A good use case for this model is when you have a new app for development or you want to run an experiment. In most cases, the workload is short-term, spikey, or unpredictable.
2. Spot
AWS has A LOT of servers and it is near impossible to have all of them in use all the time. AWS maximize on the unused compute capacity by using spot instances that offer a 90% reduced cost compared to on-demand instances. This discount comes with the requirement that you release the instance when Amazon EC2 needs the capacity back. This is known as Spot Instance interruption.
If AWS terminates the instance you are not charged for any partial hour of usage. If you terminate the instance you will be charged for every hour that it ran.
These instances are best used for applications that are more distributed, scalable, and fault-tolerant. One use case for this is in organizations that run a large number of tests every day, they could use spot instances with containers and CI/CD to improve their test response time all while reducing their execution costs.
3. Reserved Instances(RI)
This instance is designed for applications that have a predictable or steady usage. They can be shared between multiple accounts in an org and are more affordable compared to on-demand instances with a varying cost that depends on the following options that you have to choose from.
Term Length
This is the period that you will commit to using the service, with the option of 1 or 3 years.
Interestingly, unused RI’s can be sold to other AWS users in the Reserved Instance Marketplace.
Payment Options
How you decide t pay for the service will impact the total amount you will pay. The three options are All Upfront, No Upfront and Partial upfront. The more you pay upfront the greater your savings.
Class Offerings
There are three classes to RI’s Standard, Convertible and Scheduled.
Standard requires you to set and lock all attributes(i.e instance type) in advance for the period you selected. It offers up to 75% reduced pricing compared to on-demand
Convertible offers up to 54% reduced pricing compared to on-demand and allows you to scale up your attributes.
Scheduled gives you the option to reserve an instance for a specified period, for one year. i.e Weekly for two hours starting from a date given.
4. Dedicated
This is the most expensive option. Unlike all other options where you share hardware with other customers with your instance separated by virtual isolation i.e software, an Amazon EC2 Dedicated Host is a physical server fully dedicated for your use.
This option is designed to meet requirements that won’t allow sharing the same hardware as other customers. This could be for security concerns or standards set in large organizations.
This option is offered in both on-demand and reserved models.
5. Saving Plans
This is a newer flexible pricing model that allows customers to save up to 72% in exchange for making a commitment to a consistent amount of compute usage (e.g. $10/hour) for a 1 or 3 year term. With a Savings Plan, usage up to the commitment will be charged at the discounted Savings Plan rates and any usage beyond the commitment will be charged at regular On Demand rates.