May
03
2011
In what is likely to cause increased doubts in the viability of cloud-based servers and systems, the Amazon Elastic Compute Cloud (EC2) crashed last week. This in and of itself is not a bad thing. Which is to say, computer systems have crashed before. Web servers have gone down and web sites have become inaccessible. Email servers have gone down and customers have been without their electronic mail for extended periods of time as a result. In fact, in the days of network television (as in ONLY network television) broadcasters had their signal go out and the channel would go dark for a period of time. So the idea that a company might suffer an outage and we would lose signal for a time, that is not new.
However, because the EC2 is responsible for portions of an abundance of other web sites, the outage caused not a ripple of interruptions, but a tsunami of outages across the Internet. This, was something new for customers.
In fact thousands of companies use Amazon’s Web Service to run their web sites through EC2. The EC2 is hosted in five distinct regions, Northern Virginia, Northern California, Ireland, Tokyo and Singapore. Customers rent space on these interconnected availability zones and in return Amazon promises it will keep their web site up and running for 99.5 percent of the year, or they will compensate the company 10 percent of the cost of operating the site. This scenario allows for just 4.4 hours of down time each year. Following the crash last week some companies were down for more than a day and a half.
In fact, Reddit, Quora, FourSquare, HootSuite, ProPublica and portions of the New York Times web site, were among the more than 70 sites affected by the crash of EC2. This was a huge swath of popular web sites and the impact extended even further as traffic was diverted and customers began questioning why access was limited.
And why was access limited? What happened to bring down the EC2 and AWS? That’s a good question and one that Amazon has been slow to answer, citing the fact their engineers were still investigating the problem. What they have logged concerning the incident is that a networking event at the Virginia data center caused a domino effect in the region which caused many of its storage volumes to automatically create back-ups of themselves. This caused available storage space to fill up and hence led to the crash. This was not supposed to happen. In fact, Amazon had suggested to companies that because of the redundancy in their system, and the fact their services are spread among a variety of regions, something like this was very unlikely to happen. It happened anyway.
Does this spell the end of cloud computing? Does this mean cloud computing and cloud-based server farms are inherently unsafe for companies? Of course not. It means, like most car buyers learn at one time or another - buyer beware! There is no flawless system, whether it is hard wired or in the cloud. Systems crash, it is the natural order of things. These crashes lead to improved systems and better functioning in the long run. How companies handle these situations and whether or not they are prepared for the eventual crash is something else entirely. But they have been warned.
Would you like to find out more about how a Virtual Assistant works? Well visit our website and find out more on how to hire a Virtual Assistant and what they can do for you.