Font Size: a A A

The Optimization Model Of Grid Resource And Task Scheduling Strategy Based On Market Mechanisms

Posted on:2013-10-23Degree:MasterType:Thesis
Country:ChinaCandidate:N ChenFull Text:PDF
GTID:2268330425497318Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
The grid is a federal system of distributed resources based on open and general protocol. It attempts to integrate various resources as needed within the scope of global Internet, in order to achieve an effective solution of the resources effectively aggregate and user needs.Grid resource management and task scheduling is a very important research topic in the grid field. However, dynamic, heterogeneous and autonomous of the grid environment determines the traditional resource allocation and task scheduling is not suitable to resolve the issues. Market has the same grid characteristics, such as distribution, autonomous and decision-making, in addition, market competition and price adjustment help to achieve optimal benefits, which is precisely suitable for the solution of such problems.In this article, the market mechanism is used to optimize the grid resource management models and task scheduling. In the Multi-dimensional QoS constraints environment, drawing on the classic period and budget constraints DBC algorithm, researching the independent task, this paper explores multi-dimensional dynamic QoS demand preference, and then put forward our resources management optimization model and the multi-objective optimization task scheduling algorithm.The main work of this paper is as follows:First, the grid resource management model is optimized, mass heterogeneous pool of resources in accordance with the type and performance is divided to groups, forming a virtual resource group. Homogeneous resources is divided into the same group, greatly reduced the complexity of the resource management problem size. And, accelerate the speed of the task to find the available resources, improve the resources best match probability.Further, the multiple dynamic priorities of resources are defined, reflecting the user’s multi-dimensional QoS, and the MDP-DBC algorithm is proposed. In the budget and the deadline for the constraints, we chose the task completion time, the initial bid, the credit value and the cycle success deals number as the priority option. As an important parameter of the node, the credit value dynamic changes with the trading performance and time. A good credit mechanism will effectively guarantee the stability of the transaction and the expansion of the market scale.Then, the pricing trading algorithm based on the bargaining mechanism is presented. Different from the traditional bargaining mechanism, bargaining time and bargain rate constraints to the algorithm are added, which more in line with the actual situation of the transactions in the grid environment.Finally, in GridSim environment, the algorithm is simulation. Compared to other models and algorithms, the results showed that our algorithm has better performance, saving the cost of user, improving the trading performance, enhancing the resource utilization.
Keywords/Search Tags:Resource Management, Task Scheduling, Groping, MDP-DBC, QoS, BargainMechanism, GridSim
PDF Full Text Request
Related items