The Research On Key Technologies Of Fault Tolerant Web Services

Web services gains extensive concern from industrial community to academic community because it has wide foreground in the domain of e-business, application integration and business flow management, etc. In recent years, many research institutions, organizations and companies devote to the research and application of Web Services. The industrial community mainly concerns in standardization issues of specifications and protocols in Web Services. The academic community mainly researches on the underlying techniques of Web Services. At present, many key problems in Web Services are still need to be addressed. A very important one is how to ensure the availability of Web service. Fault tolerance is the key technology to improve the availability of Web service. We research on the key technologies of fault tolerant Web Services according with the characters of Web Services. The aim of our work is to provide platform support for the development and management of fault tolerant Web services.In the base of existing achievements, we delve into the following key technologies: computing model, replication algorithm, hierarchical adaptive failure detector and fault tolerance management. The contributions of this dissertation are as follows:(1) Computation model for fault tolerant Web Services. The system model is the key technology of fault tolerance. We investigate the three typical system models and propose a computation model named FTWSCM in the base of service model and reflection model. The model has following advantages: It accord with the character of Web Services; The fault tolerant mechanisms are transparent, easy to use and also flexibly customized; The fault tolerant properties are flexibly configured; The target service programmers almost needn't to care the fault-tolerant mechanisms.(2) Improvement of the performance of fault tolerant algorithm. The replication algorithm decides the redundancy mode and affects the performance of the services. It's a key issue in the domain of fault tolerance. Aim for improving the performance of active replication algorithm, we propose a new replication algorithm named RRR (Rapid Response Replication). It improves the performance of active replication algorithm without damaging the availability. Its basic idea is: all replicas receive request, but only the fastest one sends back the response to the client after it handles the request. Its main advantages are: In the algorithm, the response is sent back directly by the fastest replica after it handles the request; The algorithm avoids the redundant nested invocation problem arising by active replication); The algorithm avoids Consensus. Additionally, we propose a new method to avoid the multilevel redundant nested invocation problem.(3) Hierarchical Adaptive Failure Detector. Failure Detection is a key issue in fault tolerant computing. It's the first problem to be settled in a fault tolerant system. At present, many implementations of failure detectors are based on a suppose that the failure detector communication reliably with the monitored entity. They don't take into account the case of message loss and channels failure. Hierarchical Adaptive Failure Detector we proposed with good expansibility can identify the three different failure types including entity failure, channels failure and message loss.(4) Fault tolerance management. To continue work in despite of part failure, the fault tolerance system use the means of redundancy, failure detector and recovery. The fault tolerance management must be addressed by the fault tolerant system: replication management, logging and recovery management, etc. The fault tolerance management issues we studied are as follows:1) Replication management. The management of the redundant copies is the main issue in fault tolerance management. By learning the object group management mode from the object-oriented distributed computing platform, we propose a framework to perform the replication management including group maintenance and group property management.2) Logging and recovery management. Logging and recovery is the main means to recovery from failure. We propose a logging and recovery management framework which is not only transparent to the user but also accord with the characters of Web Services.(5) Design and implementation of a fault tolerant Web Services supporting platform. Based on the studies on the key technologies stated above and the Web Services supporting platform named StarWebService developed by National University of Defense Technology, we propose a fault tolerant Web Services supporting platform named SFTWS(Star Fault Tolerant Web Service) to support the development and management of fault tolerant Web services.To sum up, we present well-evaluated solutions in this dissertation for some key issues of fault tolerant Web Services. We believe that our contributions make a nice groundwork for future research and engineering on fault tolerant Web Services both in theory and practice.