Research On The Key Technologies Of Intelligent Network Disaster Tolerant System

IT systems have been the critical components of most business procedures. The availability and disaster recovery capability of the IT system have direct impact on the business continuity. In order to guarantee the organizations' business continuity, organizations must make their IT systems to have high availability and disaster recovery capability. In telecommunication network, some core systems play very important roles and the systems interruptions will affect a large scale customers and cause huge cost of down time. Intelligent Network (IN) for providing new services centrally is one of the core systems and its reliability is a mandatory issue that operators must consider.In the early days of the evolution of IN, fault-tolerant computer system is the chief platform. With the increase of traffic, the processing capability of fault-tolerant system has become the bottleneck of IN. High available computer system with higher processing capability has become swelling trend for the preferred platform of IN System. HA system ensures that local IN system has no single-point failure in multi-host environment and resumes working in a short time in case of the occurrence of single-point failure (e.g. power loss, network failure, etc.). But HA system cannot ensure the safe operation in case of major disasters. For example, when the environment of a system is destroyed by catastrophic events (e.g. fire and earthquake, etc), it is still required to provide the service availability without interruption. To meet the service availability requirement, real-time backup system located in remote site, called disaster tolerant system should be provided.The research work on the IN disaster tolerant system includes the disaster tolerance requirements analysis, the design of the IN disaster tolerant system architecture with high availability, high performance and low cost, and the modeling and algorithm analysis on the availability, performance and configuration. The main contributions of the research work are discussed in this thesis in detail and are summarized as follows:1) Risk analysis model and business impact factors for IN system and services.After summarized the IN characteristics based on the survey of IN functions, networking and operations, the IN risk analysis model is setup. A new method of business impact analysis from the view point of service plane and physical plane respectively is presented and the evaluation method for the importance of value added services is proposed. It is found that Intelligent Network system should support multiple levels of disaster tolerance for different value added services.2) Modeling of the relationship between the cost and the down time.The cost caused by the IN system interruption is analyzed. A model for the relationship between the cost and the system down time is setup in mathematical function which is helpful to calculate the maximum down time for a specified cost.3) IN disaster tolerance architecture with two levels safety protection mechanism.IN disaster tolerant four layers technical architecture and the generic disaster tolerant system architecture are proposed, and multiple topology scenarios are discussed and compared. The design of IN disaster tolerant system with two levels safety protection mechanism is presented. A wide-area load sharing and switchover mechanism is adopted in the design which is based on No.7 signaling network and has no front-end dispatch server used in normal cluster network. By constructing the availability model of the designed IN disaster tolerant system based on Markov Birth-Death process, it is proved that the system can meet the high availability requirement for disaster tolerance.4) Performance analysis and configuration computing method with minimum cost for IN disaster tolerant systemFor the designed IN disaster tolerant system, a model based on multiclass closed queuing network with FCFS is built and an approximate algorithm of MVA is used to find the system performance bottleneck. From the view points of service plane, logic plane and physical plane, the benchmarks of IN system are presented respectively. A configuration computing method fulfilling the constraints both of disaster tolerance availability and multi-services performance with the minimum cost is proposed. After compared with other configurations, it is proved that this method can give the lowest cost configuration of the IN disaster tolerant system.