Using S3 as a Web Server

There are two ways of using S3 as a web server:

• bucket endpoints
• website endpoints

Bucket endpoints and website endpoints #

A bucket endpoint allows direct access to S3 objects using HTTPS. AWS automatically activates this endpoint when you create an S3 bucket. For example, a file called test.txt in the bucket gojko will be available from https://gojko.s3.amazonaws.com/test.txt. Access to the bucket endpoint is controlled by IAM. When uploading a file to S3, you can make it publicly readable (as was done with test.txt), so anyone can access it using a web browser. You can also mark the file as private, so others will need a pre-signed download policy to access it. You used this approach for conversion results in the previous chapter.

A website endpoint is an optional feature of S3 that can perform some basic web workflows, such as redirecting users or showing index or error pages. This endpoint has a different URL from the bucket endpoint, usually a subdomain of s3-website-us-east-1.amazonaws.com in the us-east-1 region, or a similar service for other regions. To use a website endpoint, you need to activate it after creating the bucket. You can activate a website endpoint for an S3 bucket from the AWS Web Console, using the Static web site hosting section of the bucket properties page. Of course, it’s possible to activate it using CloudFormation as well, using the WebsiteConfiguration properties of a bucket.

Set a custom domain name #

S3 will automatically assign a domain name to a website endpoint. It’s not possible to set a custom domain name, but you can put a CDN between the users and the website endpoint and configure a custom domain name in the CDN. This is the usual approach for creating nice web domains for serverless applications.

Requests for the bucket endpoint only work with URLs matching an exact resource path, so asking for the root object (/) will not automatically show index.html. You can set up the website endpoint to send back an index file if users ask for the root object.

Another major difference between website endpoints and bucket endpoints is that website endpoints work using HTTP, while bucket endpoints work using HTTPS. This might sound like a security problem, but for real-world production usage,​ you should put a content distribution network in front of the S3 website endpoint, so the CDN provides HTTPS access to users. If you use Amazon’s CDN, then the CDN talks to the S3 website endpoint only internally in the AWS network, so the risk of man-in-the-middle attacks between the CDN and the origin isn’t that big.

You’ll need another bucket to host the files from your web-site directory. Users will expect to see a landing page when they type a domain in their web browsers, so you should create a new S3 bucket for static assets and set up a website endpoint. Another resource is added to the Resources section on the template, with the code from the following listing (indented so that WebAssetsS3Bucket aligns with other resource names, such as UploadS3Bucket).