A distributed dynamic and customized load balancing algorithm for virtual instances

Cloud computing is one of the most prominent technologies. The fundamental idea behind cloud computing is to distribute an array of computing services by unifying and scheduling a pool of computing resources, thereby minimizing the burden on the users and helping them focus on their core businesses. These computing resources are hosted on virtual hosts and distributed on-demand to the users by cloud service providers. For efficient resource utilization, systematic load balancing of incoming user traffic across virtual hosts is imperative. In the present work a Distributed Dynamic and Customized Load Balancing (ddclb) algorithm is proposed to dynamically handle the incoming user requests for the Amazon EC2 instances. It is also used to dynamically scale the EC2 instances up or down depending on load of incoming user requests on the instances and the CPU utilization of the instances. Along with it a distributed load balancer architecture is proposed to support and implement the proposed algorithm. The proposed algorithm takes into consideration certain CPU utilization metrics of the running EC2 instances for load balancing, serving the incoming requests and scaling the instances up or down.