Inclement weather had a disastrous effect on Amazon’s hosted cloud service during this past weekend. As reported by The Wall Street Journal, electrical storms on the East Coast the evening of June 27 began causing disruptions to Amazon’s cloud services shortly after 11 PM EDT on Friday, affecting customers including Netflix, Pinterest and Instagram.
Netflix and Pinterest appeared to be operating normally as of the morning of Saturday, June 28, but Instagram still appeared to be having issues, according to the Wall Street Journal. As late as Saturday afternoon, Amazon was still reporting issues with its elastic cloud compute, relational database and elastic beanstalk services. The root cause of the outage was a severe storm in Virginia which knocked out both the primary power service and backup generator Amazon uses to support its operations there.
The More Things Change …
This is not Amazon’s first experience with a cloud services outage. On June 14, 2012, Amazon experienced an interruption in cloud services at the same Virginia data center due to a cable fault in the power distribution system. Although the backup generator initially kept power on, nine minutes later one of the generators overheated, and then the secondary backup server had a breaker failure. And in April 2011, Amazon experienced its first highly publicized failure of its hosted cloud service.
Bad Weather for Cloud Services
In all instances of cloud services failure, Amazon has offered public apologies, taken steps to ensure outages do not happen again, beefed up customer service and responsiveness, offered reimbursements, etc. And while it is true that no technology works 100 percent of the time and even the most advanced and sophisticated IT services are not foolproof, it doesn’t change the fact that in little more than a year, Amazon has suffered three embarrassing and very public cloud services failures. Beyond giving Amazon a black eye, these highly publicized cloud gaffes are also starting to give cloud technology as a whole a bit of a bad reputation.
For example, in the Huffington Post today, columnist Gerry Smith wrote “the outage highlighted the tradeoffs of cloud computing” and that “the cloud is not always dependable,” although he noted “experts said that the benefits of the cloud still outweigh the risks, and that companies should focus on building applications that can respond to future disruptions.”
Cloud technology is not going away, and inevitably it will become the predominant means of hosting IT enterprises. However, Amazon needs to do whatever it takes to minimize the risks of any additional cloud services failures in the near future (as in the next couple of years), not just for its own sake, but for the sake of cloud technology in general.