For what it's worth, I have an on-prem Nexus server with a Docker repository. It has 8 cores and 16GB RAM. It has 82000 hits/day in the webserver log, though 99.9% of them transfer only a few kB, so I assume it's a metadata check and the client already has the correct version.
The same Nexus is also hosting our Maven and NodeJS repositories. That has 1,800,000 hits per day, although all but 120,000 of them return HTTP 404. (I think one of our clients has misconfigured their build server to use our Maven repository as a mirror of Maven Central, but as it's just an AWS IP I don't know who it is.)
I'm sure it's overprovisioned, but the marginal cost of 2, 4 or 8 cores and 4, 8 or 16GB RAM isn't much when we buy our own hardware.
> the marginal cost of 2, 4 or 8 cores and 4, 8 or 16GB RAM isn't much when we buy our own hardware
this is the crux of it all. RENTING bare metal (eg from hetzner) is 10x cheaper than aws ec2. So, I can only imagine how much cheaper it is when you buy the hardware directly.
At a very rough estimate with Hetzner's calculator, which doesn't have exactly the systems I most recently bought, our servers cost about 18 months of renting the closest Hetzner dedicated server.
Add to that rackspace costs, staff time doing installation (or cost of remote hands etc). I don't have our figures for rackspace costs, but staff time is fairly low — an hour or two when the server is purchased and later thrown out, and maybe the same to replace a HDD at some point in the 5-8 years we keep it.
Also add the time to get competing quotes from Dell, HP etc and decide on the configuration, but then a similar process is needed to choose from Hetzner vs their competitors.
I haven’t tried running this yet, but it seems worth keeping in mind. It’s relatively simple software so the idea could probably be pretty easily adapted to other situations.
I recall mentioned here ttl.sh, which as I looked it up [0], uses through Docker a CNCF project called Distribution Registry [1] which implements the core container registry functions (and appears to have additional utility, like being a pull-through cache).
Even if it is a daily number, it's not very much. There are 86,400 seconds in a day. Even limiting the time to an 8 hour business day, this is only around 1 pull per second.
