Big-name cloud providers like Amazon Web Services (AWS), Microsoft Azure and Office 365 guarantee a monthly uptime of 99.9 per cent. While that might sound high, in real-world terms 99.9 per cent uptime equates to a little over 43 minutes of downtime in a month, or eight hours of unplanned downtime in a year. Think about what that level of downtime could mean for your business.
A small business might be able to survive a few hours of website or email downtime, but a larger size business may lose far more. The loss in sales, customers and reputation, for instance, can become extremely damaging and costly. And remember, cloud providers will only provide a miserly discount of your monthly service bill in the event of downtime as opposed to providing compensation for the actual loss to your business, which may be far more.
You don’t have to look far for examples of businesses that have committed all of their systems to the care of cloud providers without a contingency plan.
The ferocious storms that hit New South Wales in mid-2016 knocked out the operations of numerous AWS customers including Domain, Domino’s Pizza, Foxtel, The Iconic and Stan. These businesses were left without websites or key systems for hours, crippling their ability to trade from mid-afternoon until as late as the following morning. Then there was Microsoft’s Azure outage last year, which left a number of Australian businesses cloudless for over seven hours.
These outages serve as a warning that sending business-critical systems to the cloud, rather than hosting them on-premise, does not remove the risk of costly failures on its own.
Design with the assumption of failure in the cloud
To ensure maximum availability and performance, the risk of the businesses’ cloud computing operations should ideally be spread across different regions or multiple providers. Replicating your server instance will ensure you're covered even if multiple regions or cloud service providers experience outages.
With the right tools, you can even quickly provision that replica infrastructure into a different data centre as needed, instead of sinking budget into keeping those additional resources on standby.
In a multi-zone scenario, if AWS’ Sydney data centre experiences an outage, then an affected business could maintain uptime by simply switching over to the AWS Singapore data centre, continuing to trade as normal.
While a multi-zone cloud strategy should satisfy the risk concerns for most businesses, running mission-critical applications simultaneously across multiple cloud platforms such as AWS and Azure might be an attractive option for those wanting maximum availability.
In a multi-cloud scenario, if overall access to all of Amazon’s data centres goes down, you could failover to Azure or Google. According to a recent survey conducted by RightScale, 82 per cent of enterprises have a multi-cloud strategy in place.
The bottom line is: Don’t put anything in the cloud that is mission-critical without a contingency plan for downtime, whether that's synchronisation with a back-up cloud or a series of local servers that can be switched on. Structuring businesses technology in such a way will ensure your critical applications are built to survive failures of any data centre.