AWS Reserved Instances (RI) is a term agreement for reserved capacity and resource allocation to be used by EC2/RDS/other instances in specific criteria. RI can be used when you have a dynamic fleet with rapidly changing compute requirements, to ensure that capacity reservations are met. The primary advantage includes RI's ability to float across accounts in AWS. Since multiple users can share the same RI, it cuts down costs.
How Reserved Instances work?
If you buy an annual train pass, it will cost a lot upfront, but you'll end up saving in the long run if you take the train often. RI works in the same manner.
The purchase option that you choose will determine your discount:
No Upfront, Partial Upfront, or All Upfront - The more you pay upfront, the greater your discount is.
Standard vs. Convertible - Convertible plans provide flexibility. They can be exchanged for a new instance type, new OS, different tenancy, or a different instance size. Standard plans, however, give you a greater discount.
Term - A three-year term will save you more, but a one-year term comes with less commitment.
Availability Zone (AZ) Scope vs Regional Scope RI
AZ Scope: Guaranteed capacity for the instance type that you reserve. The purchased RI would not automatically float across AZ in a Region but would be contained to one AZ.
Regional Scope: No capacity guarantee for the instance type that you reserve. You have the flexibility to apply the Reservation benefit to any instance of that type running in a particular Region.
Add customized tags to individual reserved instances in AWS, and track the cost using CloudSpend