I recommend Puma and work for Heroku, though threadafety is still a huge issue. It also pre-forks in hybrid mode which makes it super awesome. Also it takes care of slow clients!
We fire up sidekiq and clockwork in threads within the puma process rather than take ~ 100 MiB/process hit of running them separately.
If not, I would love it if you wrote one. :)
https://github.com/steakknife/rails41rc_plus_hacks_and_threa...
Feedback requested in the issue tracker. Anything that seems out of place.