Camelot is correct. You are asking for site-specific info.
I'm going to assume you're connecting remotely into your own network, and make some guesses, and hopefully help you recall how your network was set up.
What you enter depends on what access you've opened up into your network.
You can certainly try all the protocols you've listed. ftp, http, and the rest. (Well, not ftp. Avoid ftp. ftp is a pile of gnarly problems, and one of the best ways to make sure that anybody monitoring your network connections from the coffee shop will know your login credentials. On a good day — one where your password and user aren't compromised — you'll still have to battle with getting the emphemeral port range working and open on either the client end or the server end, and through all of the intervening firewalls. As should be obvious, I'm not a particular fan of ftp. Given a choice, I'd use sftp, or maybe ftp via a VPN.)
If your NAT device is correctly configured and at (more or less) its default settings, then absolutely everything you try for your remote access will be blocked.
Everything should be blocked, too, as you really don't want more than absolutely necessary open, as there are more than a few folks that would look to access your server. The folks attempting remote access will generally find and try any open port, too. Even an open port associated with a relatively secure protocol can still have vulnerabilities, whether it's a protocol weakness that somebody discovers, or a weak password.
Now if you want to enable remote access of file services or such, then your NAT box will need to have TCP port 548 (AFP) configured and opened through, or whatever ports for whatever other service you want.
Beyond being insecure, ftp is a pain in the rump to open, as you either need a device that sniffs the traffic and opens the ports necessary, or that triggers on a connection, or you'll need to have the whole of the ephemeral port range open.
More appropriate for various remote-access contexts would be a VPN, either established via what's known as NAT pass-through (mapping the VPN ports through your NAT box to your OS X Server, and configuring the VPN server on your OS X Server box), or via VPN established directly into whatever gateway box you're using, if your gateway is mid-grade or more expensive box and has an integrated VPN server. Not all of the NAT boxes have VPN support.
If you're in the Finder Connect To Server prompting:
- If AFP (TCP port 548) is already open, then afp://youraddress
- If CIFS/SMB (TCP port 445) is open, then smb://youraddress
But only somebody that's scanned your network or set up your network will know your configuration and which ports to use, both because of the NAT settings as well as any firewall you might be running on the server — OS X Server 10.6 could and variously did have the firewall service configured and running.
If you want to see what ports are already open, there are various remote services that offer that test. For instance, log into the server and use your web browser there to connect to GRC and run their Shields Up test; you'll need to run that port scan from within your network. That'll tell you which ports are open through into your system.