How do I deploy public-buckets on Railway?

You can deploy public-buckets on Railway by clicking the "Deploy Now" button on this page. Railway will automatically set up all the necessary services and configurations for you.

What are the system requirements?

Railway handles all the infrastructure requirements. You only need a web browser to deploy and manage your application.

Is this template free to use?

Yes, this template is free to use within your existing Railway account.

Can I customize this template?

Yes, you can fully customize this template. After deployment, you will have full control of what you've deployed in your Railway project.

How do I get support for this template?

You can get support through our community forum: public-buckets-541dc895, or join our Discord community for more assistance.

Deploy and Host Public Buckets

cdn-s3-go

Railway Buckets provide an excellent offering to store static files, however there are 2 major limitations:

They are not public-accessible
They are not geo distributed (region of the bucket is fixed)

public-buckets template deploys a webserver that and wraps any S3 compatible buckets and makes a CDN like access layer. It's geo distributed, generates pre-signed URLs to avoid exposing buckets to public, and most importantly - utilizes HTTP 302 redirects to return content without proxying the data.

Dependencies for public-buckets Hosting

The backbone is cdn-s3-go, a super fast, open-sourced Go service that does the job. It has very little amount of external dependencies: external deps

There is an optional Redis service, that provides a caching layer for pre-signed URLs. If you see that it negatively affects performance - just delete it.

Demo Project

There is a demo project available for this template. It serves the following URLs from demo bucket:

Each one image exists only in one regional bucket: one in US and another in EU. Guess which one is where.

Configuration

Template automatically creates an example bucket and example environment variables. There is the main gist of it:

Webserver config accepts list of bucket names as env variable CDN_BUCKET_PUBLIC_NAMES like lfs,avatars. In this template we have example.

It also accepts list of region aliases as env variable CDN_BUCKET_REGION_ALIASES like us,eu. Template deploys only us, you are free to change it.

For every bucket name in the list, it tries to load S3 / Railway Buckets compatible credentials:

CDN_BUCKET_ENDPOINT_{public_bucket_name}_{region_alias}
CDN_BUCKET_REGION_{public_bucket_name}_{region_alias}
CDN_BUCKET_NAME_{public_bucket_name}_{region_alias}
CDN_BUCKET_PATH_PREFIX_{public_bucket_name}_{region_alias}
CDN_BUCKET_ACCESS_KEY_ID_{public_bucket_name}_{region_alias}
CDN_BUCKET_SECRET_ACCESS_KEY_{public_bucket_name}_{region_alias}

Here for instance how demo project was linked:

bucket links in demo

See the full reference for variables in .env.example.

About Hosting public-buckets

Deploying a Go based docker image, that can be replicated across regions, as it's a stateless gateway. The gateway manages auth to Railway Buckets and acts like a public CDN layer. There is a Redis cache involved, but completely optional, it is included in the template by default, but can be removed in few clicks.

Common Use Cases

Hosting static files for your next SaaS, for example - avatars and customer data
Distribution of large blobs, for example, model weights or UGC content as part of the AI processing pipeline
Upd.dev uses this CDN layer for Git hosting

Deployment Dependencies

Docker image xlab/cdn-s3-go
No other external dependencies

Why Deploy public-buckets on Railway?

Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it. In this case, infinite amount of S3 compatible buckets, balanced via Metal-hosted edge layer with little help of redirect magic.

By deploying public-buckets and using cdn-s3-go on Railway, you are enabling your own CDN layer to deliver static across the globe to customers, overseas AI infra, while keeping egress costs zero.

Who needs R2 after that!?