Throttled Based VM Load Balancing for Cloud Computing Systems

Abstract

Since past decade, cloud computing secured the top rank in the list of the most needed technologies in IT industry. As the IT industry increasing rapidly, there is a huge requirement for computing, storage, data sharing and other services. This new paradigm Cloud Computing, is nothing but ”a pool of configurable resources which are readily available for the usage via any network medium on a rental basis.” Cloud provides a different set of services like storage, platforms, software, development tools, and infrastructures hosted by third party vendors. Users, who want to use a cloud no need to own any infrastructure they can simply access these services from worldwide on a demand basis and have to pay only as much as they have used.
Cloud guarantees scalability, flexibility, efficiency, and fault tolerance to its customers. Many organizations from different fields like medical, banking, research centers, and much more are moving towards cloud because of these features. Even the count of individual cloud users is also increasing. Cloud has to satisfy all these requests. Cloud owners try to accommodate a maximum number of users with the minimal cloud environment, whereas users and organization expect maximum usage of the cloud with minimal cost. With Virtualization concept, this may be possible. Load Balancing mechanism, which balances the workload in the system, become important research topic due to this need. Load balancing algorithm makes sure that no virtual machine is overloaded or under loaded so that all the resources are effectively used. With better resource utilization we can achieve our goal without compromising satisfaction of owners and users.
In the first part of this thesis, we have discussed basic building blocks of cloud, Load Balancing and some existing Load Balancing algorithms, their advantages and limitations. In the second part, we have proposed an approach for the VM Load Balancing along with vm allocation policy. We have discussed the advantages of our approach with example results. Results shown that in our approach, allocation is more fair.