Fault Tolerant High Speed Network Services

The predominant concern of evolving high-speed network architectures is fault tolerance. High-speed networks will expereince a variety of faults, such as link or node failures, excessive buffer overflow due to faulty flow and congestion control mechanisms, and partial failure of the switching fabric. The main goal of this research is to increase the availability of essential network services without compromising efficiency and cost. We proopse a modular software fault tolerance toolkit, called charon, which can be applied at the application level to make network services tolerant to faults that are not handled at the hardware and operating systems levels. Charon consists of a library of functions providing network services with heart beats, checkpointing and critical data recovering, and a mechanism consisting of four watching processing to monitor the heart beats of the netwrok services and to recover crashed processes back to thei last checkpoints. With the toolkit, network services can achieve fault tolerance with a minimum amount of changes to their source code. In this proposal, the functionality of each of the modules in Charon will be studied. The application of the toolkit to a typical network service for improving fault tolerance will be demonstrated. ANTICIPATED BENEFITS: The anticipated result of this project is a toolkit and a mechanism to achieve fault tolerance in a wide range of network services. The toolkit outlined in this proposal will have a huge market potential for software and information systems vendors who are required to provide to users fault tolerance and high availability without excessive overhead.