Welcome to the interconnected global village. A giant storm on Australia's east coast over the weekend destroyed homes in Sydney – and destroyed part of an evening's viewing and pizza-ordering for couch potatoes on this side of the Tasman.
Welcome to the interconnected global village. A giant storm on Australia's east coast over the weekend destroyed homes in Sydney – and destroyed part of an evening's viewing and pizza-ordering for couch potatoes on this side of the Tasman. Online banking was also affected.
It seems medieval but cloud computing is susceptible to stormy weather, as illustrated by Amazon Web Services' problems on Sunday, which took service offline for clients including Lightbox on this side of the Tasman.
AWS (a subsidiary of Amazon, best known for e-tail) is the world's largest cloud computing outfit.
On Sunday, its giant Sydney data centre was hit by a power failure, caused by the massive storm that stretched from Brisbane to the south coast of New South Wales.
Like all cloud operators, Amazon has its own generators, data centres in multiple locations, and triple redundancy.
But for whatever reason, the failsafes did not kick in on Sunday – or at least not for several hours. It seems that when the systems that are supposed to handle reduncancy stuff out, you're stuffed.
Problems first hit at 3.47pm local time (5.47pm NZT) and hit different clients for different durations. Most companies with services hosted by AWS suffered for two hours. But others, via social media, were still complaining about intermittent issues close to six hours later.
Kym Niblock, chief executive of Spark's Lightbox, says her company's streaming video-on-demand service was taken offline for around an hour by AWS' problems in Sydney. Full service was restored by about 8pm NZT.
Domino's online pizza-ordering service was also hit, as was Westpac's online banking.
So was Foxtel, which was streaming rugby via AWS, as was Presto, a Netflix clone owned by Foxtel and Seven.
Xero is another high-profile AWS client. However, its software-as-a-service is hosted primarily by Amazon data centres in the US. A spokeswoman tells NBR there were no reports of service interruptions. Xero started life with AWS competitor Rackspace, but is now in the midst of a two-year transition to AWS.
The AWS outage capped a bad week for streaming video that saw Sky's inhouse-managed Go service fall over, yet again, after it apparently became overloaded on the eve of Game 7 of the Oklahoma City Thunder vs Golden State Warriors NBA conference final series.
'Plan for failure'
"No cloud vendor purports to be completely infallible. Rather, they all suggest that customers 'plan for failure'," cloud computing commentator Ben Kepes says.
"In practice what that means is that organisations must ensure they have multiple points and paths of redundancy. Perhaps the best example of this in practice is, ironically, the company Lightbox is emulating, Netflix. It seems Lightbox would have been advised to talk to some people who understand cloud redundancy."
Compensation?
The whole gamut of AWS services was effected, from online storage to computing-on-demand to content delivery.
The affected services included EC2, Elastic Load Balancing, ElastiCache, Redshift, Relational Database Service, Route 53 Private DNS, CloudFormation, CloudHSM, Database Migration Service, Elastic Beanstalk and Storage Gateway.
It's unlikely any of the companies affected will get commercial compensation, however.
AWS's terms state: "AWS will use commercially reasonable efforts to make Amazon EC2 and Amazon EBS each available with a Monthly Uptime Percentage of at least 99.95%, in each case during any monthly billing cycle."
Follow NBR on Facebook, Twitter, LinkedIn and Instagram for the latest news and free on-demand audio from NBR Radio.
