Sayısal ekonomi şebeke sistemleri için yeni bir zamanlama modeli

AbstractComputational grids have emerged to exploit geographically distributed resourcessuch as clusters or idle personal computers to solve large-scale computational and datademanding scientific problems. It has been considered that developing computationalgrid economy systems in which users pay for using resources or services, would motivatepeople to share their resources making the computing power economically available thatthe communities require. In this thesis, we present a novel economic-based jobscheduling heuristic to be used in such a grid system. The heuristic basically tries tocomplete a sequential workflow or a parameter sweep application using one or moreoptimization strategies (cost, time or time-cost) according to the deadline and budgetconstraints of the user. The experimental results reveal that our heuristic outperforms therelated heuristics in the literature. Besides, we present two market models, a commoditymarket and a combinatorial double auction model, that are expected to meet therequirements of the resource owners and users in the economic respect and ensureefficient scheduling in a computational grid economy system. We performed simulationexperiments to compare the market models, and the experimental results demonstrate thatthe models have both advantages and drawbacks in terms of achieving social welfare inthe market.