Secure entry into remote systems

[Edward]

As most of you know, I cannot exactly call myself a noob anymore :(

However, when it comes to administrating remote PCs I certainly can say 
I am.

I have several servers installed at customer's premises. I used to 
simply run out there to fix any slight problems or update mail white 
lists etc.

However, with a few customers more than 1.5 hours drive away, I need to 
look at remote administration. Especially for simply adding few users to 
a spamassassin white list or the like, which really only are 10 minute jobs.

So, I was thinking about setting up dyndns or no-ip addresses for these 
servers, then opening up the firewall for either ssh or VPN. None of my 
customers have a static internet address.

I've used ssh locally before, and that is really simple to set up, but 
because of the open hole I'll be creating my question is really: Should 
I be learning about setting up VPN tunnels into their systems instead?

Anyone have any true experience using both and can shed some light on 
the security implications? Also, we're in Western Australia, with 
archaic PSTN networks (56K modem - so only 33K upstream), so any 
overheads incurred using one over the other I should also consider?

If the answer is VPN - can anyone give me a link to a tutorial or 
something to get me started? I'm by now fairly versed in Linux itself 
(Since RH5.2 anyway), have compiled kernels and the like, but VPN is new 
to me.

Regards,
Ed.