I am running a rails app with 100,000 pages views/month with a 86MB slice(not at slicehost though) with <2sec response times. The pages are pretty heavy dynamic pages.
Well, 100k/month is only 1 view every 25 seconds, so while it sounds impressive in aggregate I wouldn't like to be around if you ever get Dugg. But I'm impressed you can run anything at all in 86M really. What is that, a single mongrel instance, nginx, and mysqld?
Lets see. I have 3 slices, 2 of which are publicly accessible.
512 MB:
Apache2 / PHP, serves a dozen websites last time I checked, most notably a forum that has a few thousand hits a day and my blog, which when it gets spiky goes up to X0,000 hits in a day.
256 MB (until recently):
Bingo Card Creator, which is Nginx => 2x Mongrel.
I bumped the 256 up to 512 (a process that took me above 5 minutes and a restart) when I started developing the new version, so that I could fit a staging server on the same box. Currently it has about 6 Rails instances running on it at any given time (2x Mongrels for the live site, 2x for the staging, 2x for DelayedJob or consoles). After I release the new version I'm probably going to put the staging server on another box and reallocate the saved memory to more instances for the production site.
I have never come anywhere close to my limit. I think the highest I've ever gotten was a bit over 100 GB a month from the sites, with another 50 GB through my (private) web proxy (I have a bad Amazon tv episodes habit). My limit is 500 GB at the moment, so I'm not too worried.
