Users are not all local to your time zone. They may be spread across the globe. Not only that, but - even if they were all local - different subscribers have different usage profiles. Many office workers use their systems primarily 8-5. And software developers all night it seems :-). Retailers often use systems until 9 or 10 pm. Entertainment industry users may need their system well into the night. And realty and mortgage companies tend to use systems intensely on the weekend, when home buyers are most active. We have customers in all of those categories. This continuous system workload is another effect of both multi-tenancy and the global dispersion of the user base. There is never a good time for a maintenance window, not even on the weekend, if ‘maintenance window’ means an outage of the services. We deal with that - as do all cloud service providers - by minimizing the need for maintenance outages, so we can accomplish most updates without an outage, and only need to take the system off-line briefly in case updates need to be made to a database schema. (PS: For larger enterprises, this challenge applies in their on-premises environments as well. Only smaller enterprises operating in one or a few contiguous time zones have a workload variation with slow periods suited for a maintenance outage. In the cloud, there are no slow periods).