Research On Key Techniques Of The Trust Based Resource Scheduling In IVCE

Virtual computing environment aims to provide harmonious, credible and transparent integration service supports for applications via the public computing platform built upon the open Internet infrastructure. With the development of virtual computing technologies and the deepening of applications, on the one hand, the underlying resources gradually become open, scale and isomerization; on the other hand, the upper application requirements present a diversitified, discrepant and complicated trend. How to effectively solve the resource scheduling problems between massive as well as heterogeneous resources and diverse as well as multi-required application in the public environment, to realize the credible service target under the unpredictable and public running environment, which increasingly turns into key problems to influence and restrict the development of the virtual computing environment.This paper aims at resource scheduling problems in the public environment, combines the concrete situation of experiment platforms in the virtual computing environment. By gradually analyzing the resource properties as well as application requirements, on the basis of research results of the existing resource scheduling technologies, we investigate several key techniques based on trust resource scheduling under the virtual computing environment, in order to realize gradually fine matching and scheduling between the resources and application, thus enhance the overall performance of the virtual computing environment. In this paper, the main innovation points include:(1).We propose a redundancy scheduling algorithm based on Bayesian Trust Model--TR_Scheduling algorithm. Through the preliminary analysis of the load bearing applications on the virtual experimental platform, the applacaitons have the following features:single task is small size and large concurrency, the task are independent of each other, the task can be executed in any order, the task don’t need communication of each other, and according to the advantage of large amounts of resources in the virtual computing environment, implements redundant scheduling algorithm based on multiple copies by mean of the design concept of "appropriate resources exchanges for complex fault tolerant overhead". At the same time, according to Bayesian Trust Models, analyze the stability of schedulable resource nodes, compute the rank of resources trust value and present the maximum priority (HighFit) algorithm and the minimum priority (LowFit) algorithm for dynamic resource selection, improve system reliability and resource utilization..(2).We present a kind of resource scheduling model implementing differential service, namely resources scheduling model based on resource reserve technologies RRM model. Through the preliminary analysis of application rereqiurements, the applications, supported by platforms, can roughly be divided into "general application" and "important application" according to the important degree. RRM resources scheduling model, based on computing resource properties, put common resources and high quality resources into "common resource pools" and "VIP resource pool", respectively, and through the dynamic adjustment of proportion of VIP resources, dynamic exchange of high quality and common resources, reasonable resources reserve strategies, this model, on the basis of taking into account all the task schedulings, focuses on how to improve VIP task scheduling efficiency and implementation success rate, realizes the differentiated services of important resources servicing to the important tasks, and efficiently ensures that the application of important service level can get timely responses,improve system usability..(3).We come up with a kind of resource scheduling model realizing multiple constraint conditions, which is a kind of resource scheduling model based on "resource sliding window" techniques-RSW model. Through further analyzing application requirements, we can draw a conclusion that the application requirements, supported by platforms, own the characteristics of diversification, differentiation and complication together with multiple condition constraint for scheduling resources. This model does a preliminary study on resource clustering technologies, and makes the select range of resources gradually narrowed down through classifying the static property and dynamic properties of resources in advance, in order to realize the purpose of "appropriate resources servicing to the appropriate tasks", improve system suitability.(4).We put forward a kind of resource scheduling model based on fuzzy clustering scheduling-FCS model. By further analyzing resource clustering technologies and classifying the static properties of resources, we cluster the resources again based on fuzzy association clustering algorithm, so as to narrow the choice scope of resources and reduce the space complexity of directly clustering original resources, therefore it can provide conditions for an exact match of tasks and resources. Simultaneously, we realize optimal resource allocation "specific resources servicing to specific tasks" utilizing the scheduling algorithm, further improve system suitability.