* Run a proxy server that caches pages. Basically all software supports proxies.
Sure, basically everything supports proxies. But if you're not running a MiTM attack the proxy doesn't see much. The proxy server gets CONNECT <domain> and relays the request without ever even getting the full URI for the requested resource, so it certainly can't cache it.
* Run a HTTPS caching server and add the CA to the systems.
I think there's several challenges here:
- This CA empowered MiTM attack isn't just proxying Wikipedia and other pubic sites anymore. It's also proxying banking information, and other things that should be secure. We've meaningfully reduced the security for everyone in that lab.
- Where previously a computer playing the part of the proxy could just sit between the computers and the internet at large, allowing new machines to be added at will. Now every new computer will need some operating system & browser specific setup work. Machines that move around (laptops, rasberry pis are super portable) will end up with a variety of random certs on them.
Even with a local SSL caching server, we're still looking much slower pages. The high ping & packet loss means that the extra requests required to set up the connection could add seconds of latency to even loading a page.