Adobe was in the spotlight recently when its Adobe CS Cloud service, Creative Cloud, went offline for 48 hours, leaving users in the dark and preventing publication of the mobile edition of Britain's Daily Mail. This was a disaster for the company and a much bigger disaster for thousands of Creative Cloud users trying to meet urgent deadlines -- but in future failure in cloud services could damage the global economy.
You see, cloud services sometimes fail:
- Microsoft Azure, an essential component of that company's plans, suffered two significant outages last year, one of which left some users locked out completely and others unable to manage applications for eight hours.
- Amazon in September saw a two-hour outage at one of its data services, during which all services hosted there were unavailable.
- Dropbox suffered a 15-hour outage in January 2013.
These short outages are perhaps less significant than the failure of Adobe Creative Cloud, but they are serious problems for anyone for whom time literally is money.
Continuity software company Neverfail warns enterprise users who have become reliant on cloud services to ensure that they have backup plans for cloud infrastructure in case such problems happen. These plans should include at least some access to key data and applications -- though Adobe's 48-hour failure also meant its desktop users experienced problems.
Users shouldn't simply be wary of cloud service failures -- reliance on these services exposes them to real problems in the event their service shuts down. That's what happened to Nirvanix customers last year when the company closed shop and gave them just two weeks to move all their data out of its facilities. Customers had little time to get that data back, and no guarantee the data would be destroyed.
Cloud services are billed as being always-available solutions that enable business efficiency and flexibility for a fee -- but the need to prepare for unexpected service outages and the risk of service closure mean CIOs considering such deployments need to budget for additional costs -- not something you'll generally see in the promotional literature.
Additional cloud costs also include the need to secure data privacy and security; preserving the flexibility to move data between services and defining data nationality (where it is stored).
Enterprise customers can protect themselves by insisting that their service-level agreements include mechanisms to safeguard against outages. "This is a fundamental duty of due diligence in commercial and organizational terms," Experton analyst Alex Oppermann told CIO.com.
Cloud implementations are everywhere already. A recent 2013 IDC survey reports that 61% of enterprises already have at least one cloud-based application, and investment is up over 10% since 2012.
The impact is greater -- far greater -- than most people realize.
These days, if you go shopping at your local mall, you're probably already using cloud services, as Cloud-hosted omnichannel retailers exploit incredibly complex data analysis systems to do such things as stock control, figuring out your likes and dislikes (personalization), and delivering useful offers at checkout.
Retailers say they want to improve your shopping experience in order to make more money and deliver a unified experience online and on the street. They are becoming dependent on these systems - Britain's Tesco last year made about $45 million in pure profit by discounting prices intelligently using such systems; in the U.S., Target is investing deeply in cloud-connected intelligence.
You see, Cloud-based services are becoming increasingly entwined across the global economy -- from banking, to retail, to the B/OSS systems used by carriers and beyond. As cloud systems become more embedded in business systems, the potential damage of cloud service failure is growing exponentially. We have already reached a point at which, the Atlantic Council think tank warns, a major cloud service failure could initiate a chain of highly damaging economic events.
"If that failure cascaded to a major logistics provider or company running critical infrastructure, it could magnify a catastrophic ripple running throughout the real economy in ways difficult to understand, model or predict beforehand," the Atlantic Council's report said. "Especially if this incident coincided with another, the interaction could cause a crash or collapse of much larger scope, duration and intensity than would seem possible -- similar to the series of events that struck the financial system in 2008."
This close coupling of Cloud services with the non-virtual economy means it's only a matter of time until a serious incident takes place, they believe. "Internet failures could cascade directly to internet-connected banks, water systems, cars, medical devices, hydroelectric dams, transformers and power stations."
The Atlantic Council recommends that enterprises take steps now to protect themselves, insisting on:
- A commitment to at least 99.5% uptime. - Financial penalties for service failure need to be locked into the SLA|.
- Knowing where data is hosted and the legal requirements around that data.
- Easy data migration so they can shift to other service providers without lock-in.
- A regular "real-world" backup of all data held in the cloud.
- Rock-solid security.
Cloud services are already part of the economic infrastructure. As they become even more pervasive, the consequences of service failure become even more serious.
Certainly Adobe's massive failure caused serious problems to its customers, and its communication with customers since has been far less than satisfactory. However, the incident serves as fair warning as to the likelihood and potential severity of cloud failures. If -- or perhaps when -- such problems happen to a major infrastructure provider, the effect could threaten the entire economy. Ironically, in the name of business efficiency.