When a server is 'shut down' for phishing or spam, a firewall blocks all incoming/outgoing except for traffic to/from a pre-determined IP address along with the notice that the server is being quarantined.
Site owner/admin can then access the server, perform any investigations or deletions necessary, notify data center, then data center opens traffic again.
Alternatively, something like a web-based shell allowing access, but all other traffic denied, would be acceptable.
I've had servers shut down for 'abuse' which was one complaint from someone at 1am local time for me. I supposedly got a call from xxxxxx at 2 am, notifying me that action would be taken, and my server was taken offline at 3am. I wasn't awake until 7am, and couldn't get things resolved until about 10am. I was told I needed to 'rectify the situation', but how can I do that when access to the server is blocked? It's a ridiculous execution of policy, and only serves to heighten everyone's frustration. A private web-shell or single IP in the firewall to allow access would resolve most of the ill-feelings site owners caught in this situation have had.