Research On Trust-based Cloud Service Management Model And Scheduling Algorithms

Cloud Computing is regarded as one of the most important forms providing future web services. However, cloud has been faced with serious security problems. So to propose reasonable security solutions is currently the hottest topic in cloud researches. Practices have showed that trust is an effective alternative means of security in distributed web environment. Since cloud is also a distributed, heterogeneous and dynamic platform, the introduction of trust to cloud provided an effective way to solve cloud security issues. Although researchers have done many useful exploration works in traditional environments. Cloud’s unique features such as ultra-large-scale, resource totally virtualized and focusing on providing on-demand services make traditional trust strategies cannot be used directly. According to the problems mentioned above, this dissertation chose cloud security as our research object and mainly focused on cloud trust model and trust-based solutions including trust-based cloud service discovery model, trust-based meta-task scheduling strategy and trust-based workflow scheduling strategy. The main contributions are as follows.(1) A domain based recommendation transaction feedback cloud trust model is proposed. The new model divides cloud into several trust domains and relys on the reliable third party or domain trust agent to help cloud entities manage their trust relationships. Trust recommendations are managed as one of cloud transaction types. Recommendation transactions increase the trading times between cloud entities which accerates the establishment of trust relationships. Experimental results show that our model can help to end the early chaos and build trust relationship network more quickly compared with similar models. Besides, it has the ability to effetively resist malicious attacks and ensure high transaction success rate.(2) A trust fuzzy comprehensive evaluation based cloud service discovery model is proposed. The new model uses trust multi-dimensional vector to represent the integrity of the cloud suppliers when provide different services and use fuzzy comprehensive evaluation method to evaluate their integrated trust rating. Based on users’service demand, it has the ability to find the vantaggio providers in the corresponding transaction context. The model emphasizes cloud provider’s comprehensive service capabilities and cloud’s purpose of on-demand services, and thus obtains more accurate service quality evaluation than the traditional methods. Experimental results show that the new model can help cloud users fast and more accurately find the providers whose primary business is their demand service and thus the transaction success rate is also high.(3) A trust based resource fuzzy clustering cloud task scheduling algorithm is proposed. The new scheme divides cloud task scheduling problem into two stages:user scheduling and task scheduling. It introduces trust mechanism into the macro user scheduling level. Meanwhile, in order to reduce the scale of resource selections, it uses fuzzy clustering method to classify resources into computation, transmission and storage three types and binds user’s tasks to the resources in the corresponding resource pool according to their service preferences. The new algorithm emphasizes scheduling security. Through the classification of scheduling and resource clustering, it reduces scheduling problem’s scale and better meets cloud’s target of providing on-demand services. Theoretical analysis and simulation results show that the new model performs better in time performance and overall customer satisfaction compared to other kindred traditional solutions.(4) A trust-based customizable cloud workflow scheduling algorithm is proposed. Similar to meta-task scheduling, it divides scheduling into multi-workflow scheduling as the unit of user and micro single-workflow scheduling two stages. it adds trust factor into cloud workflow scheduling goals. In multi-workflow stage, it put forwords trust based scheduling policy. While in single-workflow scheduling stage, it analyses different workflow’s service preference and designs customizable scheduling strategies for them. Theoretical analysis and experimental results show that the new scheme achieves some advantages in scheduling effectiveness, work efficiency and user satisfaction compared to the classical models.