Hello Stack Overflow Users.
I did a speed test on CloudFront.
The environment is as follows.
S3 bucket (Tokyo): CloudFront unused 28MB
S3 bucket (Tokyo): cloudfront use 28MB
File receive command input
S3: wget -S -v S3 file url
CloudFront: wget -S -v cloudfront / file url
(We did each twice for accuracy.)
It is the average value of 2 times.
Tokyo
CloudFront not used: 1.7 seconds
Using CloudFront: 1.7 seconds
Sydney Region
CloudFront not used: 2.8 seconds
Using CloudFront: 3.0 seconds
California Area
Unused CloudFront: 30 seconds
Using CloudFront: 16 seconds
Frankfurt Area
Unused CloudFront: 65 seconds
Using CloudFront: 75 seconds
I think it would be faster to use CloudFront as a result of the California region.
But why are the Sydney and Frankfurt areas slower?
Are you having trouble with my test?
As @michael-sqlbot points out, a Miss from cloudfront should not be faster (but actually somewhat slower) than going towards S3 directly. Even when you get Hit from CloudFront CloudFront is not itself faster than S3 when served from the same location. Usually, the differences are marginal, and can be in the favor of S3.
The perks of using CloudFront when used in front of S3 are: