paradist / bytebin

Fast & lightweight content storage web service.

bytebin is a fast & lightweight content storage web service.

You can think of bytebin a bit like a pastebin, except that it accepts any kind of data (not just plain text!).
Accordingly, the name 'bytebin' is a portmanteau of "byte" (binary) and "pastebin".

bytebin is:

• fast & (somewhat) lightweight - the focus is on the speed at which HTTP requests can be handled.
  • relatively low CPU usage
  • relatively high memory usage (content is cached in memory by default, but this can be disabled)
• standalone - it's just a simple Java app that listens for HTTP requests on a given port.
• efficient - utilises compression to reduce disk usage and network load.
• flexible - supports any content type or encoding. (and CORS too!)
• easy to use - simple HTTP API and a minimal HTML frontend.

I host a public instance of bytebin for some of my own projects, which you are welcome to use too.

There is also a traditional "pastebin" frontend for sharing code/configs/whatever, see lucko/paste for more information.

Running bytebin

The easiest way to spin up a bytebin instance is using Docker. Images are automatically created and published to GitHub for each commit/release.

Assuming you're in the bytebin directory, just run:

$ docker compose up

You should then (hopefully!) be able to access the application at http://localhost:3000/.

It's that easy!

API Usage

Read

• Just send a HTTP GET request to /{key} (e.g. /aabbcc).
  • The content will be returned as-is in the response body.
  • If the content was posted using an encoding other than gzip, the requester must also "accept" it.
  • For gzip, bytebin will automatically uncompress if the client doesn't support compression.

Post

• Send a POST request to /post with the content in the request body.
  • You should also specify Content-Type and User-Agent headers, but this is not required.
• Ideally, content should be compressed with GZIP or another mechanism before being uploaded.
  • Include the Content-Encoding: <type> header if this is the case.
  • bytebin will compress server-side using gzip if no encoding is specified - but it is better (for performance reasons) if the client does this instead.
• A unique key that identifies the content will be returned. You can find it:
  • In the response Location header.
  • In the response body, encoded as JSON - {"key": "aabbcc"}.

Public Instances

I host a public instance at https://bytebin.lucko.me

You can use it in your application as long as:

• you're not malicious
• you don't needlessly spam it
• your usage isn't illegal, infringing copyright or otherwise going to get me into trouble
• you provide a User-Agent header uniquely identifying your application

If you're planning something likely to be super duper popular or use a lot of data (say >5GB per month across all users), then please run it past me first - otherwise, go for it!

If you come across any content which is illegal or infringes on copyright, please get in touch and let me know so I can remove it.

Credits

bytebin uses:

• jooby and netty for handling http requests
• caffeine for caching and rate limiting
• gson to read the configuration file on startup

and plain ol' Java for everything else.

Performance

I haven't had time to do any accurate benchmarks or performance testing, however, the libraries bytebin uses (see above) are known to be pretty efficient.

The public instance handles approx ~250k requests per day and stores ~1M items at any one time. It uses very little CPU resources and memory only up to the desired cache size.

License

MIT, have fun!