Wouldn't continuous deployment and rolling forward be better than switching your entire application architecture? In this scenario, you deploy ~30 times a day.

Wouldn't continuous deployment and rolling forward be better than switching your entire application architecture? In this scenario, you deploy ~30 times a day. If there's a bad commit, you revert it, and then do another deploy. So there's no rollbacks, you don't lose as much velocity, and since deploys are safe and a reverted commit is just a new commit, the revert is safe too.