Abstract: Parallel processing in Distributed Computing refers to the concept of running several tasks simultaneously on different processors. Load balancing and scheduling are very important tasks to optimally utilize the available resources and processor utilization. These problems are NP-Complete. In this paper, we introduce two methods which are genetic algorithms and Honey bee algorithms for scheduling and load balancing in parallel heterogeneous multi-processor systems. Genetic algorithm is a meta-heuristic algorithm which provides optimization by performing genetic operations on the running tasks and Honey bee algorithm is a Bio-inspired algorithm which provides optimization based on foraging behaviour of honey bees. Finally these two types of algorithms provides better optimization than the normal CPU scheduling algorithms, it is studied by comparing the results of meta-heuristic and bio-inspired algorithms with CPU scheduling algorithms, such as Longest Processing Time (LPT) and Shortest Processing Time (SPT) . The results of simulations indicate meta-heuristic and bio-inspired algorithms for scheduling and load balancing provides better total response time and system utilization. Simulations results indicate Genetic Algorithm and Honey bee algorithm reduces overall response time and also increases resource utilization in load balancing of distributed systems.
Keywords: Heterogeneous multiprocessor systems, Meta-heuristic algorithms, Bio-inspired algorithms, Genetic algorithms, Honey bee algorithm, Waggle dance, Virtual machine, System utilization, Total response time.