Load Balancing in the Cloud Using Specialization

On the Internet of the future, billions of physical objects will collect and exchange data that will be stored in the Cloud Computing. A bad storage structure of big data has a negative influence on the latency time during the client's requests. To solve this problem, we have proposed an Infrastructure as a Service (IaaS) for the input and the output of different files on servers' clusters. We have proposed a specialization based architecture and used a multi-agents system paradigm. Some proposed agents collaborate to realize the load balancing, while others work in a parallel way to achieve their tasks rapidly and minimize the latency time. We implemented our solution using the JADE platform. We measured the execution time and the response time for each request type (input/output) and for the file type (text, image, and video). By comparing the execution time and the response time using our architecture and the classic one, our architecture has given better results.