I wrote about this a few months ago: https://rythie.com/blog/blog/2014/03/05/should-all-sites-use...
It is technically encrypted in HTTPS traffic but it isn't treated with very much respect so if you actually have access to all of the HTTP and DNS traffic surrounding a request you can often recover pages viewed.
Additionally, in a lot of these countries computers come pre-installed with a government root CA which they can use to impersonate sites like Wikipedia (although the USG does this too!).
"A user agent MUST NOT send a Referer header field in an unsecured HTTP request if the referring page was received with a secure protocol." http://tools.ietf.org/html/rfc7231#section-5.5.2
and that's how browsers implement it too.
DNS doesn't give the page you were on. Whilst some systems might have a government root CA on it, it's still quite possible to remove that - it's pratically impossible to remove ISP level monitoring.
Google, from 2010: "On our production frontend machines, SSL/TLS accounts for less than 1% of the CPU load, less than 10KB of memory per connection and less than 2% of network overhead."
https://www.imperialviolet.org/2010/06/25/overclocking-ssl.h...
Entropy is a different matter, but I believe pretty much all virtualisation platforms have ways to ensure the VMs have enough entropy sources - so it should be fine.
As for entropy, your server only needs a small amount of entropy to seed a CSPRNG, and the CSPRNG takes it from there.