aws-region-proximity

AWS Region Proximity

Print a JSON dictionary where the key is an AWS region code name (us-west-2, eu-central-1, etc.), and the value is a list of regions in order of their physical proximity to the region. You may optionally pass in a list of region code names that you have a copy of the resource in. The results will have the same keys, but the lists will only contain the regions you specified.

Why?

Say you have S3 buckets being cross-region-replicated in us-west-2, eu-central-1, and ap-southeast-1 and you want to put a Cloudfront distribution in front of them. You can use a Lambda@Edge with an Origin Request trigger to change the origin that Cloudfront will pull a resource from. Lambda@Edge has the environment variable AWS_REGION, which will tell you the closest region to where your Lambda is executing. You could then use the list generated by this program to help find the nearest origin that can handle the request.

For example, a request comes in to the DFW edge location. The environment variable AWS_REGION is "us-east-1" (the AWS region associated with the edge location of DFW). You don't have a copy of the resource in us-east-1, but you do have a copy in us-west-2. So that can be the copy you return (as opposed to your copy in eu-central-1 or ap-southeast-1).

Install

pip install aws-region-proximity

Usage

Get the Closest Regions for All Regions

aws-regions

Output:

{
    "us-east-1": [
        "us-east-1",
        "us-east-2",
        "ca-central-1",
        "us-west-2",
        "us-west-1",
        "eu-west-1",
        "eu-west-2",
        "eu-west-3",
        "eu-central-1",
        "eu-north-1",
        "eu-south-1",
        "sa-east-1",
        "ap-northeast-1",
        "me-south-1",
        "cn-north-1",
        "ap-northeast-2",
        "cn-northwest-1",
        "af-south-1",
        "ap-south-1",
        "ap-east-1",
        "ap-southeast-1",
        "ap-southeast-2"
    ],
    ...
    "cn-northwest-1": [
        "cn-northwest-1",
        "cn-north-1",
        "ap-east-1",
        "ap-northeast-2",
        "ap-northeast-1",
        "ap-south-1",
        "ap-southeast-1",
        "me-south-1",
        "eu-north-1",
        "eu-central-1",
        "eu-south-1",
        "eu-west-2",
        "eu-west-3",
        "eu-west-1",
        "ap-southeast-2",
        "us-west-2",
        "us-west-1",
        "ca-central-1",
        "us-east-2",
        "us-east-1",
        "af-south-1",
        "sa-east-1"
    ]
}

Get Closest Specified Regions per Region

For when you have copies in a limited number of regions and just want to see your closest available region given some other region.

aws-regions us-west-2 eu-central-1 ap-southeast-1

Output

{
    "us-east-1": [
        "us-west-2",
        "eu-central-1",
        "ap-southeast-1"
    ],
    "us-east-2": [
        "us-west-2",
        "eu-central-1",
        "ap-southeast-1"
    ],
    "us-west-1": [
        "us-west-2",
        "eu-central-1",
        "ap-southeast-1"
    ],
    "us-west-2": [
        "us-west-2",
        "eu-central-1",
        "ap-southeast-1"
    ],
    "eu-west-1": [
        "eu-central-1",
        "us-west-2",
        "ap-southeast-1"
    ],
    ...
    "cn-north-1": [
        "ap-southeast-1",
        "eu-central-1",
        "us-west-2"
    ],
    "cn-northwest-1": [
        "ap-southeast-1",
        "eu-central-1",
        "us-west-2"
    ]
}

There is still a key for every region, but the list of the closest regions is limited to those specified on the command line.

From Whence Comes This Data?

This Gist: https://gist.github.com/atyachin/a011edf76df66c5aa1eac0cdca412ea9

Instead of geographic coordinates, it'd probably be better if we sorted by actual inter-region latency, like what this table has: https://www.cloudping.co/grid

Put it on the TODO list.