Oversized compute config leads to significantly higher spends.
AWS has a huge number of options when it comes to EC2 instance types. The same is true of Azure VM’s also. As on date there are 600+ EC2 instance types to choose from
As a Startup or Small business, it is critical that you select the right server config. Else you end up incurring high costs or face performance issues.
Both are detrimental to business. Also, the same applies to other compute resources like RDS , Kubernetes Clusters , ECS ,etc
In this post I talk about EC2 instances . But the same approach can be successfully applied to other AWS services (RDS, EKS, etc) as well as for Azure systems and service.
How to choose the right EC2 instance type for your app , website or backend ?
There are only two key variables. vCPU and Memory. Still choosing the right combination from 350+ EC2 instances is confusing.
Challenges faced by Startups and Small businesses in selecting the right AWS EC2 instance type are :
- Very little to no visibility about the load pattern on EC2 instance
- Actual resource utilized per user is not clear. Since user engagement with application is evolving.
- User base is scaling or not fixed
- Limited finances. Can’t afford to spend more than actual need
How does one go about selecting the right AWS EC2 instance type? What is the approach for deciding the correct EC2 instance type ?
You can spend time and efforts working out the projected capacity and the sizing, but it ends up being a futile exercise. Based on my experience working with Startups , I suggest the below approach. It is simple , highly effective and also ensures that you costs are the least.
es, there are many permutations and combinations possible in vCPU and Memory. Yes, there can be that ideal sized AWS EC2 instance type for your need. But you can figure it out only after many weeks or months of usage.
My simple suggestion is this. Dont get too hooked up in choosing the right AWS EC2 instance for your app. Don’t spend too much time or efforts on it.
Go ahead and choose the AWS EC2 instance type you guess is closer to your requirement and budget. Start small. The trick is to monitor the chosen EC2 instance type from Day-1. If performance becomes an issue then you can upgrade to the next EC2 instance type config instantly or vice versa.
Continue to monitor for couple of weeks and use that learning to revise your selection of EC2 instance type.If you need to know other ways to reduce AWS costs or resource wastages check this out.
Core value proposition of AWS Cloud is scaling up or down on demand. We should not spend time and energies in capacity planning to decide the right EC2 instance type. Else we are using cloud like a traditional IT and not like true cloud computing.