Will Natural Disasters Kill the Cloud?
Slashdot recently posted an interesting infographic questioning whether a hurricane or other extreme weather conditions could take down cloud computing and a significant percentage of the internet as a whole.
It is true Amazon AWS (Amazon Web Services) recently experienced cloud computing disruptions and a short cloud service outage. In June of 2012, Amazon AWS customers were impacted when lightening struck causing a regional power failure. As a result of the Amazon AWS cloud computing outage, several high profile sites like Netflix, Pinterest, Reddit, and Instagram were impacted with several hours of down time. You can imagine how even a short cloud outage could dramatically impact this powerful foursome.
- Netflix: +27 million members, over 800 devices stream from their service.
- Pinterest: Hosts roughly 3.5 million active users daily, signed up 13 million users in 10 short months (2,702% increase from May 2011 through February 2012) and 180% annual company growth.
- Reddit: +39 million unique monthly visitors, influence spans across 172 countries, powers thousands of active communities daily, over 1.6 million logged in redditors.
- Instagram: +80 million global users, 58 photos are uploaded every second, more than 5 million photos are uploaded daily, over a billion photos have been uploaded to Instagram since inception.
This freak accident aside, could bad weather conditions or a huge storm really take down cloud computing and the internet? Anything is possible, but this scenario is highly unlikely.
The infographic does make a good point in it’s reference to Rackspace who:
- Is speculated to hold about 10% of the market in comparison to Amazon’s 70% market share
- Store data for over 180k businesses
- Supports 40% of Fortune 100 Companies
A disruption to services provided by Rackspace, Amazon AWS, Microsoft Azure, Oracle, and even Salesforce.com could have catastrophic wide-stream impacts, but this is nothing new. Having all of your eggs in a highly concentrated basket is scary for some, but I would rather take my chances with a top-notch data center vs. a homegrown self-managed data center any day.
Think about homegrown data centers and lets use the financial services industry as an example. These guys may be able to mount hardware, run cable, hire fancy electricians to load balance power, setup a few redundant systems, and ensure gigantic air conditioners pump ice cold air into their spotless server rooms, but they are still data center wannabees at the end of the day. They do not eat and breathe this stuff; their bread and butter focus is on business services and products, not technology and server rooms, even if they make every attempt to follow industry standards and disaster recovery plans (DR).
Now think about state-of-the-art data centers. Their entire business model focuses on providing the highest quality environment for hosting data. All company investments and business priorities are aligned to continual improve and perfect the data center experience for their customers. Entire departments are dedicated to business continuation planning, disaster recovery (plans for earthquakes, hurricanes, lightning strikes, torrential rain, and even meteor strikes), data security, facility management, engineering, and the list goes on. Companies like Rackspace eat and breathe everything surrounding the hosting of data. Their livelihood is dependent on it. They do not have unrelated competing priorities. Secure and protected data is their only focus.
Additionally, the world’s top data centers in fact do not place all of their eggs in the same basket. Redundant data center sites setup in separate and distinct disaster zones is a commonly used practice. For example, a company would not have two of it’s data centers located a few miles apart. Rarely are they even contained in the same state. Usually, the two data centers would be strategically positioned to ensure they do not share the same central power grid, seismic zone, and even network backbones. This usually means data center A is located on the West coast while data center B is located on the East coast. In an event where a hurricane or other natural disaster takes out the primary data center, the backup site can quickly and seamlessly step in to ensure there is little to no disruption.
In my opinion, the accountability needs to be shared for any company who hosts their data externally. Outsourcing makes sense, but it is their responsibility to ensure the data center meets their highest standards. They should conduct a physical site review and scrutinize every detail as if they were building the data center on premise themselves. If having both data centers in the United States is “cutting it close”, their solution should focus on data centers who have backups both in the states as well as another country or continent.
How I look at it, if there is a storm large enough to take down the internet, meaning primary and backup sites are simultaneously knocked out, we won’t have to worry about it anyway. World economies would collapse and the “fittest” will survive and weather out the storm… pun intended.
What are your thoughts? Is it possible for a silly hurricane to bring cloud computing and some of the largest companies to their knees, or is this infographic more dramatic than realistic?