Ip Network Service Survivability Model And Mechanism Studies

IP networks have demonstrated their great importance in the development of society. However a tiny failure or fault would make it fail to work efficiently. Naturally, it is one important issue to enhance IP network survivability, i.e., to make IP services more survivable under adverse conditions, such as network attracks, natural disasters, human element accidents, etc. With such background, the thesis does some innovative research on the survivability of IP network services.(1) Considering that the existing service migration methods bring the server cluster heavy burden, and their time spending is so high that seriously affects the service continuity. We put forward an intent-perceptible service migration model, called ISM. In this model, we design a novel trigger mechanism and object selecting method for service migration, and make the service migration intention perceptible for the client with the cooperation of the server. The evaluation results show that ISM model can not only ensure the service invulnerability by maintaining a high randomness of service migration, but also improve the service continuity by reducing the service gap time spending on service migration to the maximum extent.(2)We propose another service migration model, called CPSM. The model empowers the client to select servers freely according to its own preference and to migrate a service from one provider to another proactively when the service is threatened by attacks or failures. We believe that our proposal is the first of its kind to integrate several well-established theoretical and practical techniques from survivability, operations research, and network security to form a client-side service migration model. The principal component of CPSM is a service migration controller located at the client side which abstracts service migration as a Markov decision process, and it is able to compute the optimal service migration strategy in the case of network failures or intrusions by observing services'states and evaluating servers'reliability. We provide the motivation and theoretical fundaments for CPSM, and expatiate on its architecture and implementation details. Then through some simulations, it is proven that CPSM can maintain service continuity at the user-perceived level under adverse conditions, and also shorten the average response time of services. Thus, CPSM is of guiding significance to survivability research of IP network services.The thesis mainly has three innovative points as follow.(1) Improving of the traditional trigger strategies and object selection methods in service migration;(2) Realization of sensing function by the client towards the service migration intention inside the server cluster;(3) Construction of the decision maker for service migration at the client side which entitles the client to implement service migration suiting itself.