#zerovm

Van Lindberg here from Rackspace. If you have any questions, I am around to answer.

This is a very insightful Q & A into the architecture, philosophy and potential of ZeroVM. One of the main developers for ZeroVM is also present and provides some informative answers.

I've been looking into ZeroVM as a means of breaking down a traditional request-serving app that executes on only one machine into a finely-grained, short-lived, 'spawning like crazy' sandboxed lightweight environment that could run off a single hypervisor . The 'one instance per user per app' paradigm is very intriguing and could lead to much better utilisation of existing hardware (which is great for people like me who have to squeeze every cycle out of the limited, existing hardware available to them).

I installed ZeroVM quickly and painlessly today thanks to the simple, working documentation on their website (lets see OpenStack achieve that.. ok, ok, OpenStack is quite a different beast, I'll admit that - settle down). I'm playing with it inside of a virtual environment but ultimately would like to see it run on the bare-metal of my dell (cheap, second-hand, loud!) poweredge 2650.

Most of my software development takes place with Ubuntu machines running python (Django, Flask, etc) so my first task after having become familiar with the orchestration of ZeroVM instances is to port a simple web app, maybe a hand-rolled Flask blog or a basic Django website, to make use of on-the-fly instantiated ZeroVM instances. The main thing to bear in mind at first will be changes to the development and debugging cycle, and how the architecture or development flow will need to be adapted to accommodate temporal server instances.