Page 168 - How Cloud Computing Is Transforming Business and Why You Cant Afford to Be Left Behind
P. 168

MANA GEMENT STRATEGIES F O R THE CL OUD R EV OL UTION



                 Health Dashboard. Amazon regularly reports on the status of
                 its services through the dashboard. It uses a little green icon
                 to indicate that a service is functioning normally or a yellow
                 one to indicate a problem. Normally, the dashboard is littered
                 with rows of continuous green symbols.
                     On December 9, at 4:08 a.m. Eastern time, Amazon posted

                 a warning symbol with information that indicated, “We are in-
                 vestigating connectivity issues for instances in the US-East-1 re-
                 gion.” Such information is frustratingly nonspecific to the
                 cloud user. But there was more. Eighteen minutes later came
                 the notice, “We are experiencing power issues ...inasingle
                 availability zone in the US-East-1 region.” What had hap-
                 pened, according to subsequent notices, was that a primary

                 power supply component had failed. “Prior to completing the
                 repair of this unit, a second component, used to assure redun-
                 dant power paths, failed as well, resulting in a portion of the
                 servers in that availability zone losing power,” Amazon Web
                 Services posted to its dashboard.
                     These notices give the EC2 cloud a degree of transparency,
                 a prized attribute in cloud operations, but if an individual cus-
                 tomer depended on them alone, his view into his virtual ma-
                 chines would be far from transparent—somewhere between

                 translucent and opaque.
                     In fact, the user needs to maintain some independent view
                 into cloud operations to know whether her virtual machines
                 are running or stalled. As it turns out, a new service, Apparent
                 Networks’ PathViewCloud, had been in operation for about a
                 month, and it tracked the service outage to a router on the





                    148
   163   164   165   166   167   168   169   170   171   172   173