Our vision is that to achieve true elasticity new instances should be spawned for each incoming request. We have put together a small demo to demonstrate that a new-instance-per-request approach is feasible.