We have a Windows 7 machine with a cell phone hotspot connected to the Internet (DHCP on the hotspot's local network). We run OpenVPN and TightVNC. The hotspot's signal or latency seems to vary during the day/night, and that seems to have a huge effect on the performance of the VNC connection.
Further, we find that different flavors of VNC seem to have very different performance under these changing conditions. For instance, TightVNC becomes crippled, consistently disconnecting part-way through the first screen paint, while UltraVNC (which let's you throttle colors) does a little better--but not a lot. The join.me service seems to do far better (I realize it doesn't go through OpenVPN, and it has obvious limitations, but the fact that it works when others fail tells me we can do better).
Is the problem with OpenVPN or VNC, or the combination?
(a) Are there ways to tune OpenVPN for better performance under these conditions? (b) Are there ways to measure OpenVPN performance (beside ping round-trip times) and predict when it's getting worse?
If the VNC traffic is going through OpenVPN, why does the flavor of VNC matter?
Are there ways or flavors of VNC that are optimized for this scenario? I'm familiar with UltraVNC that lets you throttle down the colors for the sake of bandwidth, but are there other/better adjustments?
Should we bail on OpenVPN and try a lighter-weight transport (maybe SSH--if it is lighter weight), or some more direct method (like dynamic-DNS)?
Which question(s) am I neglecting to ask here?
We're not sure exactly what's changing in our conditions, that's causing such a drastic change in performance.
Windows 7 Pro SP1. OpenVPN 2.3. TightVNC 2.6. UltraVNC 1.1.9.6.