We've spent a lot of time over the last decade making NBD a faster and better protocol: pipelined requests, full support for trim and zeroing, load balancing over multiple connections, a proper specification, new client and server implementations (libnbd, nbdkit), standard command line tooling, interop testing between all the implementations, and a formal URI specification. It's very trivial to set up an NBD server and there's tons of documentation about how to connect to it with different clients (start here: https://gitlab.com/nbdkit/nbdkit https://libguestfs.org/nbdkit.1.html#SEE-ALSO).

Unfortunately in this specific case (diskless booting) dracut support still sucks, which I really need to get around to fixing at some point.

Well, iSCSI is a standard, so chances are better that it's supported in a non-Linux OS, e.g. MS Windows. Years ago I booted a Windows (7, iirc) client that way, but gave up on it (too much hassle and performance limited by the network) when SSDs became cheap.

I don’t have direct experience, but when I looked into it my takeaway that NBD was unable to reliably deal with network interruptions as well as iscsi.

https://forums.gentoo.org/viewtopic.php?p=4895771&sid=f9b7ac...

https://github.com/NetworkBlockDevice/nbd/issues/93

Whether that’s the case with the latest version, I don’t know, but it’s something you might test if you choose to try it.

You might like https://smolbsd.org/

Well yes, I do like that:), but I don't see the connection to this thread?

I think you need NBD if you're going to use glustre, but I could also be thinking of ceph.