If you're not picturing the difference — say you have one m5.large server on on-demand, you'd pay $840.96 over a year compared to:
If you go Reserved, the price could drop to $235.47
If Savings Plan, it could get down to $285.93
But these numbers obviously aren't fixed, they depend on when you reserve, the region, and so on.
Reserved Instances (RIs)?
Allow you to reserve EC2 capacity for 1 or 3 years in exchange for a discount of up to 72% compared to On-Demand pricing.
🔹 Types of Reserved Instances:
Standard RIs: The highest discount, but you have to be committed to a specific instance type, meaning you're confident about the load that'll be on that instance for the whole long period you're reserving it for. And if you want to get rid of them, you'll have to list them for sale on the RIs Marketplace
Convertible RIs: Lower discount than the one above, but gives you the chance to switch between instance types, on condition that the final cost ends up equal to or higher than what was originally agreed
What Are AWS Savings Plans?
More flexible alternative to Reserved Instances, offering up to 72% savings in exchange for a commitment to a minimum spend per hour for 1 or 3 years.
🔹 Types of Savings Plans:
Compute Savings Plans: Let the discount apply across more than one service — Lambda, EC2, Fargate
EC2 Instance Savings Plans Lock you into a specific family and region, but you can change the instance type within the same family freely
If we notice, we'll find that both options let you change the instance type you're using, so why do we say the savings plan is more flexible?
Reserved is basically a promise you make to AWS, so even if the plan type was Convertible RIs, even if you wanted to change the server type you'd have to send an exchange request that's manual and wait for it to get approved so the discount applies to it, otherwise you'll get charged on-demand.
The Savings Plan is more automatic because it focuses on the amount you spend more than what you're actually using, so if you change the instance or the service, e.g. from ec2 to Fargate, it'll look at your spending from, say, the last hour, and start checking what's supported, and if the discount applies to it, it'll save you more.
What do you pick?
✅ Reserved Instances if:
✔️ Your load is predictable during the plan period so you won't need to change the instance much.
✔️ You'll only use EC2 instances
✅ Savings Plans if:
✔️ You'll be using something from EC2, Fargate, and Lambda
✔️ You need the discount to focus on your spending, not what you're actually using
(1) Question: Since Compute Savings Plans has a more flexible option, does that make it better than Reserved?
No, because yes you have flexibility, but you might not need it if you know your load is constant or that you don't need anything beyond EC2 — in that case RIs would be more suitable and the discount reaches 72% instead of 66%
(2) Question: So if I don't need other services and I have two options with the same 72% discount, which are EC2 Instance Savings Plans and RIs, what do I choose based on?
The idea of the Savings Plan is that you become flexible, yes, but if there's a specific server type you know you need in a specific zone that might not be available later — say if the servers were stopped for a while and you started them again, or you're doing scaling —
Then the most suitable option is Zonal-Reserved, because it'll always keep the capacity you requested set aside for you until you need it
🤍 I would appreciate If we connect together on LinkedIn : Visit
Top comments (0)