A systematic literature review on soft computing techniques in cloud load balancing network

Providing, an on-demand facility in the cloud network is one of the finest services for cloud users. To maintain this dynamic and foremost service, a cloud network must pose the best load balancing techniques. One of the major research problems in the cloud environment is to manage the load dynamically. Load balancing issues are NP-hard (Nondeterministic Polynomial time) problems, and it is highly important to solve these problems in a large domain of cloud network to provide seamless and uninterruptable cloud services to their customers. But solving these issues demands standard computational paradigms techniques which embark the performance of load balancer. In this paper, an in-depth investigation of the literature on cloud load balancing techniques based on computational paradigms methods is studied. The investigation focuses on the objective to find how reliable are these techniques to achieve a balanced load in the dynamic cloud environment. An in-depth analysis of research articles that are based on the application of soft computing paradigm techniques over cloud load balancing published between 2009 and 2022 are highlighted. In the first part of the paper, the various load balancing methods as per the soft computing based paradigms are classified. Secondly, load balancing at VM and PM levels based on Machine Learning (supervised and unsupervised), Neural network, Fuzzy system, and Bio-inspired soft computing methods are categorized and the nature of work is evaluated. Detailed limitations are identified highlighting the improvement of research challenges using soft computing techniques in load balancing. This in-depth review will be supportive for researchers and professionals to choose appropriate learning and optimization techniques to achieve optimal load balancing solutions in the dynamic cloud environment.