The Research On QoS Based Grid Workflow Scheduling Models And Algorithms

Grid is a dynamic and open environment which integrates all kinds of distributed resources into a huge platform to achieve resource sharing and parallel computing, while, such resources may include high-performance computers, databases, remote equipments or communication apparatus. Gird workflows are of great significance while some large grid applications have complex logic process involving multiple steps and resources, even more there are temporal contacts between each process. The QoS(Quality of Service) is key factor in grid workflow scheduling which is used to measure the resource performance and satisfaction indicated degrees.Researching on the QoS based grid workflow scheduling problem, Several algorithm are proposed in this paper referring to scheduling models of various Qos representation.1. In the economic grid environment with market mode, time and cost are two main QoS requirements while researching grid workflow problems considering cost objective with one constraint. Genetic algorithm if fit for solving combinatorial optimization problems but it sometimes has degradation and prematurity phenomenon, immune algorithm and genetic algorithm is combined in this paper to solve the problem, The experimental result shows this method can outperform genetic algorithm and find best solution.2. With respect to the grid workflow scheduling problems for multiple QoS objectives while satisfying constraint, each objective in the problem has conflict with others and improving one objective can leads others'weaken, so actually results are usually denoted as a set of Pareto non-inferior solutions. Differential evolutionary algorithm known as its ability to solve multi-objective problems is proposed to combine into immune and genetic algorithm in the paper, The experimental result indicates the approach mentioned can achieve optimal results.3. Using heterogeneous data to represent QoS of grid resources with real numbers, interval numbers and fuzzy numbers in view of uncertainty of grid environment, mixed and multi-attributes decision making approach based on grey relational analysis is proposed to evaluate the comprehensive Quality of Services with heterogeneous data types. Such method as described is adopted in grid workflow scheduling to sort candidate services and guide the choices. Effectiveness of results can be proved in the experiment.