A Dynamic Predictive Approach to Identify an Optimal Cloud Availability Zone with Maximum Satisfaction Level
Published: 2016
Author(s) Name: Santhosh R., Ramya A. |
Author(s) Affiliation: Dept of Comp Sc. & Engg, Faculty of Engg, Karpagam University, Coimbatore, Tamil Nadu, India
Locked
Subscribed
Available for All
Abstract
Cloud infrastructure service provider allows the users to place their business application into availability zone across various regions world-wide by enabling enterprises like amazon with sufficient capability. Amazon EC2 is an instance to be placed in a zone to provision their resources to run their own applications.
Multiple availability zones are composed and
located in single region each one is different from its character specification because of its hardware and software built version. These zones allows achieving higher availability with lower failure rates but may results in different quality of service against user requirements. This situation may not advertised by cloud infrastructure service provider. In this paper, we proposed the prediction model build by repeated k-fold cross validation, which overcomes the prediction error Type-I exist in k-fold cross validation. This proposed technique helps us to predict optimum availability zone with best user satisfaction level for amazons EC2 placement in heterogeneous environment.
Keywords: Availability Zone, Machine Learning Technique, Repeated k-fold Cross Validation, Prediction Model
View PDF