AWS Cost Optimisation
Home » Cloud  »  AWS Cost Optimisation
AWS Cost Optimisation
This month we took vector to cloud solutions. Based on it I start to prepare relative topics and found that the cloud cost optimization theme does not have a lot of materials, on my opinion, because the main difficulty is that each project has its own architecture and tasks. So the advice which works for one, but will not work for another. Based on this you cant expect from this article a silver bullet solution, what you can expect is a checklist of what should you check or count for your case

Here I will explain how to work with AWS.


General:

  • Try to make data calculation inside AWS, to pay less for outbound traffic
  • Select right-sizing of the service
  • From a cost optimization point better to create as many instances as needed inside one region (same rule work for EC2 as well as for RDS instances)
  • Use VPC Endpoint service cheaper then NAT Gateway
  • Prices depend on region. When I wrote it the cheapest was in Virginia region and Sao Paulo is the most expensive
  • ALB cheaper then ELB
  • Use scheduling for your not production resources. Start and stop resources as per your needs (For example stop staging server for night time and weekend)
  • Use Amazon Elastic Container Service (ECS) or Amazon Elastic Kubernetes Service (EKS) instead of EC2 instances when you have microservise architecture: ECS and EKS can be more cost-effective as they allow you to run containers instead of instances.

S3:

  • Use infrequent access plan when you need store files which you need access rarely (some old media files or backups)
  • Use S3 as static webhost instead of EC2
  • Use CloudFront above S3 its cheaper then get data from S3 and provide caching (there no payment for sending data from S3 to Cloud Front)

EC2:

  • Use Spot instance if you have tasks that need big amount of resources and they should not work in real-time and will be not critical to run this task several times(scraping, heavy data transformation, and calculation)
  • Use reserved instances they have big discount (by 72 %) if compare with “on demand”, plus if there prepayment: more payment on advanced - more discount
  • Use Amazon EC2 Instance Store. For temporary data storage EC2 can be more cost-effective than Amazon EBS
  • Use EC2 Auto Scaling to automatically scale your EC2 instances up or down based on demand, it will help with resource utilization and reducing costs

I think these are the most common items that I use in my practice. I will try to have this list up to date, when new cases will come

Left comments, it's will be interesting to know what are you use in your practice.