There are three types of billing options with AWS :
- On-demand billing – which means you pay as you consume
- Reserved instance – which means you pay upfront for a fixed period
- Spot instances – allow you to bid on spare EC2 instance.
This post is to compare the first two billing options which i.e. On-demand and Reserved Instance. Let’s assume that
Here is the cost comparison a Linux instance with 100GB EBS storage and 250 GB bandwidth in US region
The difference is just $102 per year. So there is a saving of $8.56.
Following things need to be kept in mind :
- As you can see from the table, the discount is only in Compute, where as there is no discount in any other component
- You can scale up within the same instance family, but not in a different family. And that too only if you have UNIX/LINUX based instances. That means a Linux based m4.xlarge instance can be upgraded to m4.2xlarge but not c4.xlarge. And when ever
- You cannot change the Region of your reserved instance. A RI once purchased in Singapore, will remain there till the contract terminates
- There is no refund of the one time fees in case you want to cancel the instances
AWS RI is a very interesting way of billing the customers. Within RI you have three option of payments:
- All upfront – in which you pay the entire money of Compute reserouces upfront
- No Upfront – if you have a good billing history, you end up paying nothing and the charges are adjusted in the hourly rates
- Partial Upfront – you pay some part of the payment and remaining payment is adjusted with hourly charges
You should use Reserved instances ONLY when you are certain that you will stick to the same configuration till the end of the period.