Monday, December 31, 2012

Amazon Apologizes for Christmas Eve Outage



Amazon.com has apologised for a December 24 disruption in its cloud-computing services that hindered Netflix customers from watching movies, and said it is taking steps to prevent a recurrence.

Netflix said last week that many users in the Americas were unable to access online content on Christmas Eve because of an outage caused by Amazon's web storage and computing system. Amazon didn't identify Netflix in its statement, which was posted online on December 29, according to Tera Randall, a spokeswoman for Amazon Web Services. 

"We want to apologise," Seattle-based Amazon said. "We know how critical our services are to our customers' businesses, and we know this disruption came at an inopportune time for some of our customers." 

Amazon, the largest online retailer, has billed its 6-year- old cloud-computing offering as a cheap and safe way for customers to outsource their data centres. It's a business that may bring in about $US1.5 billion ($1.4 billion) in revenue for Amazon in 2012, according to Colin Sebastian, an analyst at Robert W. Baird & Co. in San Francisco. Occasional outages can undermine confidence in the service, which makes it possible for companies like Netflix to run their operations via the internet.

While Amazon Web Services accounts for a fraction of the $US62 billion ($59.6 billion) in sales analysts are projecting for the company in 2012, the business is more profitable than the retail operations. The segment has about a 10 per cent operating margin, Sebastian said. That compares with a 1 per cent margin expected in 2012 for the entire company, according to the average of analysts' estimates compiled by Bloomberg.

The disruption began at at 12:24 pm Seattle time on December 24 and continued until the following day, according to Amazon. The problem was related to Amazon's Elastic Load Balancing service, which apportions demand for computing power among the servers that are most able to handle the work loads.
Amazon said that data was accidentally deleted in a maintenance process run by a small number of developers, which prompted high latency and error rates for technology that manages the load balancers. At the peak of the event, 6.8 per cent of ELB load balancers were affected, the company said.

The outage occurred in the eastern US, where Amazon has data centres in northern Virginia. That region was hit by storms earlier this year that caused Netflix to lose control over streaming content for about three hours on June 29.

Amazon Web Services is used by companies in more than 190 countries, letting them rent servers for their websites, support mobile applications and store customer data, according to Amazon.

No comments:

Post a Comment