Whilst using lambda@edge your code runs in any one of the available data centres, which is not necessarily true for your data or other services perhaps.
For example, if I'm using Dynamodb global tables and my data exists in
eu-west-1
, us-east-2
and ap-southeast-1
, when my lambda@edge function
runs and sets up a DocumentClient using process.env.AWS_REGION
, the value of
this environment variable may be any of the available AWS regions, perhaps
eu-central-1
.
What this means is that the DocumentClient will fail to retrieve the data as it
attempts to access it from Dynamodb at eu-central-1
as it hasn't been
replicated to that region.
Rather than hard coding or chosing one of the three regions in which I've
stored my data, using this package I can determine which of the regions which
contains my data is closest to process.env.AWS_REGION
(or at least has the
lowest latency).
npm install @benswinburne/closest-aws-region
var closestAWSRegion = require('@benswinburne/closest-aws-region')
import closestAWSRegion from '@benswinburne/closest-aws-region'
// AWS_REGION = eu-central-1
const region = findClosestAWSRegion(process.env.AWS_REGION, {
filters: ['eu-west-1', 'us-east-1', 'ap-southeast-1'],
}); // eu-west-1
Find the closest region to that specified in to
.
Note: If options.filters
is not set, the closest region will always be to
.
The default region to select in the event that no region is found. This
defaults to us-east-1
if not set.
Specify the region(s) to limit your search to.
// AWS_REGION = eu-central-1
const region = findClosestAWSRegion(process.env.AWS_REGION, {
filters: ['eu-west-1', 'us-east-1']
}); // eu-west-1
If for example you are looking for the closest region to eu-west-1, with no
filters applied obviously eu-west-1 is the closted region. By excluding to
,
the return value will be the closest region to eu-west-1 which is not eu-west-1.
This option allows for overwriting the static datasource. The format must follow the structure used by cloudping.co.
Note: it's not recommended that a remote datasource is used with lambda@edge because the latency savings by finding the closest region are lost during the roundtrip to an external source
// AWS_REGION = eu-central-1
const response = await fetch('https://api.cloudping.co/averages');
const data = await response.json();
const region = findClosestAWSRegion(process.env.AWS_REGION, {
data: data
});