所属 コンピュータサイエンス学部 コンピュータサイエンス学科 職種 専任講師
|標題||Derivation of Optimal Number of Computational Resources in MapReduce|
|掲載誌名||Services Transactions on Big Data|
|著者・共著者||Htway Htway Hlaing, Hidehiro Kanemitsu, Tatsuo Nakajima, and Hidenori Nakazato|
|概要||Resource provisioning in cloud computing is a challenge for both cloud service providers and
users. Cloud service provides resources as a platform for the users to process large scale datasets.
MapReduce is commonly used by the users to compute large datasets from sources such as social
networking, Internet of Things (IoT), and scientific applications. A user defines the resource
requirements to run a job in the cloud and, the cloud service provider invokes the virtual machines
for the user upon request. Insufficient allocation of resources can result in a higher cost for
cloud service providers and users. Therefore, an optimal and automatic resource provisioning
approach is necessary. This paper presents the derivation of a mathematical model to estimate
the optimal number of mappers and reducers based on the specification of resources and the size
of the datasets for the automatic resource provisioning. In this paper, experimental results and
analytical results are described to confirm the analysis.