After running WP Super Cache for a good number of years, the site (AndroidPolice.com) has started to outgrow its VPS setup (5mln monthly visitors on a 2GB web + 1GB db Linode).
I've been working on introducing a 2nd box into the equation, which brought a lot of things to consider the longer I thought about various implementations, and after consulting with Donncha, we both reached a conclusion that WPSC isn't good for >1 web servers. After hearing a lot of good things about W3TC and mostly because of its memcached support, I have migrated to a dual-memcached + W3TC setup today.
So far things are going pretty well, but I did want to point out one thing that wasn't immediately apparent during initial testing and only came out later. Furthermore, with the same settings, WPSC is not susceptible to this issue.
If the option "Don't cache pages for logged in users" is unchecked, logged out users (i.e. visitors) will see pages where the comment author info field will be replaced with my name, or whoever was logged in at the time the page was cached. This is a very sneaky side effect, and WPSC somehow manages to get around this (I suspect it forces cache off even though the setting to not cache for logged in users is turned off) and definitely never caches pages in logged in state for regular visitors.
Frederick, I think you should consider putting up a more prominent warning next to that setting, as unchecking it may introduce sneaky bugs, which the site admin may not be aware of, just like I wasn't. Maybe as a workaround, logged in users could get their own cache salted with their username, but that will probably result in complications you don't want to deal with and increase the caching footprint if you have a lot of logged in users.
Thoughts on this?