When I was developing Harmony 3 last year, a point arrived in the development right before it went into beta, where I needed to perform a live test of the code. It's important to me that we are able to testify to our beta testers that the code has been tested under live conditions. At that point, the Harmony 3 code had not been tested in production, so how did we mitigate the risk of using an untested application in a live environment where everyone in the room is relying on the software to work?
Our solution is very simple, if a little tedious. On the left is a MacBook Air running an alpha of Harmony 3. On the right is a MacBook Pro running the then current release of Harmony 2.5. Both have the same service pre-programmed and both are connected to the same projector. If Harmony 3 had crashed or failed in a significant way, I would have switched the projector to the MacBook Pro and continued with service. Only later would I have taken the time to investigate any issues.
Fortunately, Harmony 3 alpha only had minor issues, which were soon corrected before it went into beta for wider testing.