I used to work for a university High Performance Computing Lab. This was one of the things I loved about it. We were 3 guys running the whole system (10k cores) and we somehow slipped between the cracks of all departments and were left to do our own thing. When a professor or student would request something, we could make the change live while still on the phone with them. While we did try to put all features into our ticket tracker at least for records sake, it was nice not having to go through a big approval process to make the changes.
This is the exact opposite of my experience working on the USCMs Tier 1 team at Fermilab (we took data directly from the CMS detector at the LHC, processed it, and distributed it across the world).
We were a team of 10 people, constantly bogged down in red tape. Oh! You want iptables rules on your thousands of nodes that are publicly facing on a Class B to the world with no firewalls, etc in the way? That's 8 months of back and forth with PHDs, committees, etc.
Needless to say, I left after a year and moved to someplace where I could effect much more change in a more rapid pace.
Man, that sucks. Places like Fermilab are perpetually starved for funding and need to get the most bang from every buck. Doesn't sound like that's happening, with processes like that in place.
