I think that your calculations are still a little bias...
Being on a VPS you are bound to have different results everytime you run the tests, as it is still a shared environment.
Something important to note is that Apache needs to be correctly configured in order to run at its best. A fully optimised and correctly configured apache should run as fast as any litespeed (providing same server, load, etc...)
I have run tests on both before, same datacenters, same servers, same connectivity and saw very little difference.
I have to agree about configuring apache but its not the easiest to configure I have no bias at all
You are right about a vps being a shared environment but I have to say with apache over the last few days the test results have been fairly similar. 5 ish seconds at 50 users however at 10-30 the speed was an acceptable 2-3 seconds per page.
However that is the 1st test I have performed with litespeed and its good so far
one thing with apache is memory usage and processor load this is a hell of a lot less with litespeed apache would run between 3.5 and 5 % however litespeed seems to max at 0.81%
Varnish is great,when i tested it on loadimpact the results where amazing
the only concern i had was the ram caching (running out of ram)
You can limit the amount of memory that Varnish can consume on a server. Be it 8M, 9M, 21M, or 2.6G. If it needs more RAM than it's allocated, it will starting moving unused objects to disk (similar to swapping).