MSFT has been selling "zero down-time patching" in 2016 for a while now, but I'm having trouble finding details at this point on the subject.
I have heard that this requires HA in your farm. This indicates to me that "zero downtime" just means that users will be pushed to an available server while the patching-in-progress server is unavailable. That's something you can do right now in 2013 with HA.
Has anyone performed this in 2016 to see the results of the proposed zero downtime?