Assessing The Costs of Cloud Outages

According to the International Working Group on Cloud Computing Resiliency (IWGCR), the total downtime of 13 well-known cloud services since 2007 amounts to 568 hours, which has an economic impact of around $71.7 million dollars.

In IWGCR’s research, they have stated that the 7.5 hours per year average unavailability only amounts to 99.9 percent average uptime, disappointing figures that are too far away from the expected reliability of mission critical systems, which is 99.999 percent. The IWGCR provides electricity in a modern capital as a point of comparison. According to their research, the average service unavailability for electricity is less than 15 minutes per year.

This is the first time that the research group published their Availability Ranking of World Cloud Computing (ARWC), since their inception in March 2012 via Paris 13 University and Telecom ParisTech. According to the group, now that more and more government agencies and global businesses are adopting cloud computing, it becomes imperative that the underlying cloud infrastructures be reliable, especially since they are being used for mission critical tasks and operations. They further added that in its infancy, the industry is not yet aware of the cloud’s lack of reliability.

The IWGCR’s research is based on press reports of cloud outages, which are usually found at services such as Facebook, Twitter, Microsoft, Amazon, Yahoo, Google, and Paypal.

The estimated financial damage of an outage that lasts for an hour can vary between $89,000, which is a figure associated with a travel service provider such as Amadeus, to a staggering $225,000 for highly critical services such as Paypal. Said figures are computed based on the industry’s accepted hourly costs. Additionally, it’s been said that large tech companies such as Microsoft, Google, and Amazon have losses that are estimated to $200,000 for every hour long outage.

The IWGCR also states that the economic impact is not the only thing that should be looked at, since downtimes that can last for days or weeks also affect millions of end users.

The researchers have admitted that their methodology is still not foolproof, due to the fact that their information gathering process was not exhaustive, but they said that the preliminary figures are actually underestimated, since information about a lot of outages are not made public, which means they may have missed a lot of outages in their figures.

The methodology used by the IWGCR has other shortcomings, such as its failure to account for the exact value of the economic cost for every outage, or the average hourly cost for each cloud service provider. More importantly, the group states that they would have preferred if the data was based on the number of users a cloud has. The group plans to adopt new methods for their next research, in order to provide a more accurate assessment of cloud availability in the future.