Ask HN: How to distribute hundreds of terrabytes of data across servers?

I've found myself in a difficult situation. I have some kind of file sharing websites that grew unexpectedly fast and I can't seem to find any good option for file distribution.

I'm using bare metal servers from Hetzner, because AWS/Azure/Google Cloud would be too expensive.

Currently I have two ways of distributing the files: 1. GlusterFS, which doesn't seem very stable, is really slow and is using a ton of resources, but it made it pretty easy and fast to setup. I don't think this is going to be a viable solution as my data is growing.

2. The second thing I tried after getting disappointed by GlusterFS was application-level distribution. What I do basically is split every HDD into 1TB partitions (so I can mirror them easier,as I'm using different size HDDs) and upload the file randomly/to the emptiest one.

The second option works pretty good, but there are some drawbacks. Searching for files is easy, but very slow with GlusterFS, harder, but faster with the second option.

My question is: What better option do I have? Did I misconfigure GlusterFS or is it normal to be slow?

Thanks!