How to deal with application downtime

Meta just suffered two major outages in a week: What can organisations do to ensure their customers aren’t confronted with service outages?