Reply to thread

Not sure if you've had any experience with actually using it yet, but how well does it scale with parallel commits and executions? Is there a new VM created for every single task/pipeline?