Cloud resource management using 3Vs of Internet of Big data streams

Internet of things (IoT) allows various smart devices to get connected to anything, anywhere, and at anytime. The ubiquitous nature of IoT devices generates huge volume of data called Internet of Big data (IoBd). IoBd is generated in continuous streams and at unprecedented speed. The rapid analysis of such IoBd streams is the need of hour. Moreover, the allocation of optimal number of cloud resources for real time analysis of IoBd streams is a challenging task. Most of the current methods use data characteristics provided by the user to allocate cloud nodes. But in case of IoBd streams, data characteristics are usually unknown to the user because of the stochastic nature of IoT devices. This poses difficulty in selecting appropriate cloud resources. This paper proposes an efficient method to tackle this issue. The proposed method first predicts the data characteristics of IoBd stream in terms of volume, velocity and variety (3Vs). Later, these predicted values are expressed in terms of a triplet called Charactrization of Stream (CoSt). On the other hand, self-organizing maps are used to create dynamic clusters of cloud resources. One of the clusters is allocated to IoBd stream based upon its CoSt. Experimental results show that the proposed method effectively boosted the performance of cloud resources and minimized the execution and waiting time of IoBd stream processing.