Investigating regions - March 2023
AWS pricing is inconsistent across services. For example, in Feb 2023, S3 pricing is incomplete for South Africa and some services price list include the China region while most do not.
Data valid on the 4 March 2023
Intro
Regions are important for AWS pricing as each region has a price list; the same product in different regions can (and often does) have a different price.
Finding some ways to compare prices per region
S3 is, with EC2, one of the oldest (mature) services of AWS. One Gigabyte of S3 storage is delivered in the say way in any region. I assume that differences in price between regions are due to the regions’ specific factors, for example real-estate prices, energy costs, and probably taxes.
Now that I think I found a good the unit of measure, let’s see how much the price changes per region.
You can access monthly updated data and graphs on regions by going to the Regions page. From there, at the bottom of the page, you can deep dive into each region.
It seems that multiple regions have exacly the same price, as if there were buckets of regions. Let’s see if this grouping is the same when using the price of one hour of t3.micro (one of the simplest instance type, available everywhere).
No! The prices are distributed differently.
I am surprise by GovCloud East 1 (and 2). It is one of the costliest region for storage, but kind of average for t3.micro; maybe because in GovCloud the data needs higher levels of admin and protection over what is normally available in the commercial cloud, while running a VM does not have such overhead.
Not all regions are created equal (in term of services)
The other important topic about regions is that not every region has every service, some regions have more services than other.
There is a mistake in the AWS pricing files: LAX is not an AWS region, yet it is returned as one.
Are there regions with all the services?
None, us-east-1 does not have it all!
My assumption that us-east-1 has everything is proven wrong. Based on the pricing data, us-east-1 does not have:
- Device Farm
- Ground Station
- IoT 1-click
- Honeycode
Services available in all regions
- Certificate Manager
- CloudFormation
- CloudTrail
- CodeDeploy
- Config
- Database Migration service…strange
- Events
- Lambda. It feels like AWS is using old EC2 hardware for Lambda
- SQS
- Secret Manager
- System Manager
- X-ray (!!!)
- API Gateway
- CloudWatch
- DynamoDB
- EC2
- Data Transfer (obvious)
- ELB
- VPC
- Elastic Container Registry - ECR
- ECF, what is that?
- EFS
- ES (Elastic Search I guess)
- ElastiCache
- Kinesis
- RDS
- RedShift
- S3, including Glacier Deep Archive
- SNS
- SWF
- States
- Elastic Map Reduce (that’s a surprise)
- KMS
Sustainability
The third important element is how green is a datacentre? Unfortunatly the data is not easily available, but I will search and see how to include being green as part of this page.