trafficserver VS Seaweed File System

in this code repository: https://github.com/apache/trafficserver

The LARGE majority of CDNs use either Apache Traffic Server (https://trafficserver.apache.org/) or Nginx for their cache webserver, so the mechanisms used are pretty easy to find if you look through the docs.

Apache Traffic Server (no relation to Apache itself) would be an excellent option: https://trafficserver.apache.org/

Although haproxy and nginx cover (for me) almost all use-cases I had to deal with (with OpenResty [1] as a backup), I see one place where ATS could shine: plugins. From examples [2], C API looks sane and well documented, and this is very important if you want to add some custom stuff inside your proxy server without losing your hair. And no, lua isn't the solution here ;)

Those who had to deal with nginx plugins, I feel your pain...

[1] https://openresty.org/en/

[2] https://github.com/apache/trafficserver/tree/master/example/...

FYI: Planning to move from github.com/chrislusf/seaweedfs to github.com/seaweedfs/seaweedfs in the coming days. It may cause some problem for package reference, building, documents, and links. Sorry for the change!

Besides storage itself, S3 API access cost can be high if frequently accessed. And latency is unpredicatble.

You can use SeaweedFS Remote Object Store Gateway to cache S3 (or any S3 API compatible vendors) to local servers, and access them at local network speed, and asynchronously sync back to S3.

https://github.com/chrislusf/seaweedfs/wiki/Gateway-to-Remot...

If you are looking at MinIO you might find SeaweedFS interesting as well.

I believe what you want is a horizontally scalable object store with tiered storage. SeaweedFS is free / open source https://github.com/chrislusf/seaweedfs

I don't know if it fits all of your requests, but you can take a look at seaweedfs, which is pretty good

Seaweedfs deserves a mention here for comparison as well.