Failure of several days at Amazon, which presented Friday apologized to customers of its data management services, has rekindled fears that causes the increase of cloud computing , whose security relies on the reliability of huge servers. “We know how important our services are essential to the activities of our customers and we will do everything we can to learn from this event and use it to improve our services,” the company said in an Amazon Web Services Message posted Friday, concluding 25 pages of highly technical explanations.
“The trigger was a configuration change,” said Amazon Web Services, a subsidiary of retail giant online for several years leases its servers at cheap price to many societies. “We will review our processes and increasing automation to prevent this error from happening again,” promised the group.
A computer security firm SANS Internet Storm Center, Johannes Ullrich, chief information officer, was pleased to learn that the failure was due to a trivial configuration issue, in short, a human error: “It is neither equipment or security. ” The advantage is that such errors can be corrected, the disadvantage is that the sector appears to proceed by trial and error. “What this shows, what are the limits of operational experience to Amazon or others to operate these systems ‘cloud’,” says Ullrich.
Cloud computing, in English the “cloud”, is for a company to use the computing power and storage available on the Internet to reduce its spending on computer equipment. This sector now generates billions of dollars in sales. The result is that hosts like Amazon or Microsoft or IBM are at once crucial for the functioning of their clients and a failure of the “cloud” can be crippling.
Several influential U.S. websites, including Foursquare (LBS), Quora (search engine) and Reddit (social network information) and had technical problems from 21 April due to this failure, focusing on servers Amazon located in Virginia (east). Amazon has not only offered to compensate its customers with a credit of ten days, but also ensured that several causes of failure were identified, “which gives us many opportunities to protect the service to prevent a similar incident happen again. ”
The problem, says Ullrich is that all this proves that cloud computing is “a very immature technology, and (many operational difficulties were not resolved.” Only the “experience” will allow to end the problems of the system, he believes. What mark a halt in the rise of paperless computer? not sure. For the independent analyst Rob Enderle, “it will just remind people that the ‘cloud’ is not bulletproof, and it takes twice systems (procedures), or with another data system, or in systems specific to the company, “he said .
For its part, Amazon has promised to improve not only its technical service, but also its communication with its customers. It took him three days, until April 24 to correct nearly 99% of the fault. Finally, 0.07% of the capacity of the system “could not be reliably restored,” admitted the group Friday.