By using AWS re:Post, you agree to the AWS re:Post Terms of Use
AWS re:Postre:Post

Please explain instance reservation recommendations

0

Can someone help me to understand the instance reservation recommendations provided by the AWS Cost Management?

For instance, we make use of the following EC2 instances:

2x m5.large 3x t2.small 1x t3.medium

But, the reservation recommendations are:

Buy 2 m5.large Buy 12 t2.nano Buy 8 t3.nano

Why? Why wouldn't I just buy the exact number and type of instances that we use? Where do these t2.nano (12!) and t3.nano (8) recommendations come from? Supposedly this is based on the last 30 days of our actual usage, but our usage has been exactly as above for quite a long time.

We also use Lambda functions.... but I don't think reservations can cover Lambda usage, right?

Similarly odd recommendations are for our RDS and OpenSearch usage -- the recommendations only partially match our actual usage.

And to be clear -- our actual usage has been stable for a long time.

Any ideas?

Topics
Compute
Tags
Reserved Instances
Language
English
AWS-User-1252302
asked a year ago540 views
1 Answer
1
Accepted Answer

There's another question which perhaps covers the "why" here. When using Linux instances you'll see that the pricing is linear - a 2xlarge instance is twice the price of a large instance (as an example). So what you purchase is normalised behind the scenes for EC2.

You're correct - reserved Instances don't cover Lambda but you might instead consider Savings Plans where you do the same sort of things as with Reserved Instances (commit to a specific amount of usage) but instead it is spread across whatever EC2 instances you use and other services such as Lambda.

For RDS and OpenSearch, Reserved Instances are the way to go - if in doubt, please contact your local AWS team and they can help.

profile pictureAWS
EXPERT
Brettski-AWS
answered a year ago
  • AWS-User-1252302
    a year ago

    Ah, I see! So for example, my 1x t3.medium == 8x t3.nano. I guess the 8x is suggested to provide flexibility.

    Is this support for equivalence new (ish)? The last time I purchased reserved instances was about 2014, and I'm pretty sure then that one has to purchase the exact same reserved instance type as was used.

  • Nataliya_G EXPERT
    a year ago

    We've launched size-flexible RIs since then (any Linux/UNIX RI with Regional scope and Shared tenancy is size-flexbile). In this case, any RI within the family and generation can apply to a running instance of any size within the same family and generation (if other attributes also match... like OS, tenancy, region...) - and RI recommendations engine will always go with the lowest size in the family to offer this flexibility to cover a certain number of normalized units (your total RI footprint), instead of exact number of instances that you are running, thus potentially increasing coverage.

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions

Relevant content