Is It Time To Get Rid Of The Linux OS Model In The Cloud? (2012)

Its certainly worthwhile to experiment with alternatives, but the various levels solve a wide array of problems -- perhaps not in the most efficient way for the cloud, to be sure -- and its not going to be time to get rid of the current model until there's a new model that adequately solves all of those problems that remain relevant in the cloud and provides performance benefits sufficient to outweight the costs of the model not be consistent with the model used elsewhere in computing (reducing, e.g., the utility of shared tooling and experience.)

I think that will probably happen fairly quickly for specialized alternative models for narrow domains (it probably already has -- I think I've seen that there are some in use now, and I'm sure there's more that I haven't heard about that are being used internally in private clouds by actors that don't talk much about what they do inside), but a general purpose replacement for the status quo style of OS stack for the cloud environment seems farther off.

Sounds a lot like the direction Erlang on Xen (http://erlangonxen.org/) is taking.