The data skew problem in MapReduce has been studied
only recently [12], [13], [14], [15], [16]. Among the solutions
proposed, some are specific to a particular type of applications, some require a pre-sample of the input data, and
some cannot preserve the total ordered result as the applications require. To make matters more complicated, the computing environment for MapReduce in the real world can
be heterogeneous as well—multiple generations of hardware are likely to co-exist in the same data center [17].
When MapReduce runs in a virtualized cloud computing
environment such as Amazon EC2 [18], the computing and
The data skew problem in MapReduce has been studiedonly recently [12], [13], [14], [15], [16]. Among the solutionsproposed, some are specific to a particular type of applications, some require a pre-sample of the input data, andsome cannot preserve the total ordered result as the applications require. To make matters more complicated, the computing environment for MapReduce in the real world canbe heterogeneous as well—multiple generations of hardware are likely to co-exist in the same data center [17].When MapReduce runs in a virtualized cloud computingenvironment such as Amazon EC2 [18], the computing and
การแปล กรุณารอสักครู่..
