Cloud services are by definition remote, so you have to be able to connect to them over a network connection. There are multiple potential points of failure, including issues with your home network, your Internet service provider, the cloud provider's network or even their own ISP, not to mention obvious things like your power going out or your computer dying. If any one of these things goes down, you are out of luck getting to your data or applications.
Network congestion will also affect how quickly and easily you can upload or download via the cloud. Extreme slowness can be just as bad as an outage when you're trying in vain to get to data you need right away to finish that last-minute presentation or paper.
Sometimes catastrophic disruptions occur over entire data centers, or portions of data centers. Services can be interrupted by a weather event or other natural disaster, a power outage, equipment failure or simply unscheduled downtime for repairs or maintenance. While cloud hosts strive for 100 percent uptime, this is a lofty goal. Things happen, even to the big players.
Amazon, an early pioneer in the cloud service area, has had several highly publicized outages from 2011 to 2013. Some of them took down or impaired well-known sites including Reddit, Netflix, Coursera, Foursquare, Instagram, Pinterest, Flipboard, Vine and even Amazon itself. They have been attributed to thunder storms, power outages, human error and even a glitch in a single networking device in one case.
Microsoft's Windows Azure service had an approximately 10-hour disruption in February 2012 caused by a leap day-related software bug. Google had an outage of only a few minutes in August 2013 that took out all of their services, including the search engine, Gmail and YouTube, and reportedly cost them around half a million dollars [source: Bloomberg]. According to the analytics firm GoSquared, the outage also reduced global Web traffic by 40 percent [source: Bloomberg, Mack].
For a person at home trying to watch a movie on Netflix, outages can be an annoyance. But for a business, even short disruptions can have a major impact on revenue. It's a lesson that anyone relying on the cloud should spread their service over multiple data centers, or even multiple providers, if possible. Backing up data regularly in some automated fashion is also a good idea. Redundancy is key. It comes with a price tag, but it may be worth it if you need constant access to your data and services.