Setting aside the other comments, I think it depends on your definition of “easier”.
Yes - it is easier in terms of up-front investment to have a scheduled (daily, weekly) downtime. But that often means the downtime is not automated - it’s someone doing the work, often out of hours when they’re not fresh, things can do wrong, etc. And it also means that the duration of the work is limited to the scheduled downtime. Some large upgrades just won’t fit in the allotted time, and you need to schedule an outage.
On the other hand, creating a system designed for zero downtime (ie, in-place upgrades, removal of big batch jobs etc) is a lot more of an investment up front, and makes upgrade planning more complex, but ultimately it saves time and effort because you can do all the changes during working hours, and “upgrades” become just how you deploy code - part of the core skill set of your team - and you do them 10 times a day.
The main difference that I’ve seen is that the scheduled downtime approach tends to create a much more conservative upgrade cycle, much bigger upgrades, and more risk of an upgrade failing - which means that a pile of features don’t get released. Also, certain customers will push back and demand a ton of documentation and detailed release notes and advanced warning of the upgrades etc - this creates loads of unproductive work for a small company. A continuous deployment model tends to reduce these problems because you’re not externalising your upgrade schedule and you don’t have a single big upgrade that can go bang, so maybe one feature doesn’t get released but everything else does.
Having taken both approaches (and even moved a company from scheduled to continuous) I would never go back to a scheduled downtime approach.