A Christmas Eve glitch traced to Amazon.com that shuttered Netflix for users from Canada to South America highlights the risks that companies take when they move their datacenter operations to the cloud.
While the high-profile failure - at least the third this year - may cause some Amazon Web Services customers to consider alternatives, it is unlikely to severely hurt a fast-growing business for the cloud-computing pioneer that got into the sector in 2006 and has historically experienced few outages.
"The benefits still outweigh the risks," said Global Equities Research analyst Trip Chowdhry.
"When it comes to the cloud, Amazon (AMZN.O) has got it right."
The latest service failure comes at a critical time for Amazon, which is betting that AWS can become a significant profit generator even if the economy continues to stagnate. Moreover, it is increasingly targeting larger corporate clients that have traditionally shied away from moving critical applications onto AWS.
AWS, which Amazon started more than six years ago, provides data storage, computing power and other technology services from remote locations that group thousands of servers across areas than can span whole football fields. Their early investment made it a pioneer in what is now known as cloud computing.
Executives said last month at an Amazon conference in Las Vegas they could envision the division, which lists Pinterest, Shazam and Spotify among its fast-growing clients, becoming its biggest business, outpacing even its online retail juggernaut.
Evercore analyst Ken Sena expects AWS revenue to jump 45 percent a year, from about $2 billion this year to $20 billion in 2018.
The service has boomed because it is cheap, relatively easy to use, and can be shut off, scaled back or ramped up quickly depending on companies' needs. As the longest-running player in the game, Amazon now boasts the widest array of datacenter products and services, plus a broader stable of clients than rivals like Google Inc (GOOG.O), Rackspace Inc RAX.N and Salesforce.com Inc (CRM.N).
Outages such as the one that took down Netflix and other websites on the eve of one of the biggest U.S. holidays are part and parcel of the nascent business, analysts say. Moreover, outages have been a problem long before the age of cloud computing, with glitches within corporate datacenters and telecommunications hubs triggering myriad service disruptions.
COMING SOON: POST-MORTEM
Amazon's latest service failure comes months after two high-profile outages that hit Netflix and other popular websites such as photo-sharing service Instagram and Pinterest. Industry executives, however, say its downtimes tend to attract more attention because of its outsized market footprint.
Netflix (NFLX.O) - which CEO Reed Hastings said relies on AWS for 95 percent of its datacenter needs - would not comment on whether they were pondering alternatives. Analysts say the video streaming giant is unlikely to try a large-scale switch, partly because all cloud providers experience outages.
"Despite a steady stream of these service outages, the demand for cloud services offered by AWS, Google, etc. continues to escalate because these services are still reliable enough to satisfy customer expectations," said Jeff Kaplan, managing director of consultancy ThinkStrategies Inc.
"They offer cost-savings and elasticities that are too attractive for companies to ignore."
But "Netflix and other organizations which rely on AWS will have to reexamine how they configure their services and allocate their service requirements across multiple providers to mitigate over-dependency and risks."
AWS spokeswoman Rena Lunak said the outage was traced to a problem affecting customers at its oldest data center, run out of northern Virginia, which was linked also to the June failure.
The latest glitch involved a service known as Elastic Load Balancing, which automatically allocates incoming Web traffic across multiple servers in order to boost the performance of a website. She declined to provide further details about the outage, saying the company would be publishing a full post-mortem within days.
AWS has traditionally been used by start-up tech companies and smaller businesses that anticipate rapid growth in online traffic but are unwilling or unable to shell out on IT equipment and management upfront.
The company has more recently started winning more and more business from larger corporations. It has also set up a unit that caters to government agencies.
Regardless, Amazon's clientele would do well not to put all their eggs in one basket, analysts say.
"Service outages do occur, but they are not common enough to cause users of these services to abandon today's Cloud service providers at significant rates. In fact, every major Cloud service provider has experienced outages," Kaplan said.
"Therefore, organizations that rely on these services are putting backup and recovery systems and protocols in place to mitigate the risks of future outages."
(Additional reporting; editing by Edwin Chan and Richard Chang)