Research on Fault Tolerance in Hybrid P2P-based Collaborative Systems

Hybrid P2P technique recently has become more and more popular for collaborative systems. However, reliability is issue of major importance in the server end of these systems. TMR technique for Collaborative System (CS-TMR) is presented in this paper to improve the server reliability. This software schema incorporates three homogeneous microcomputers and provides the fault-tolerant function through package interfaces to server applications. As it is COS-based, the method is more general-purpose, and programmers need not pay too much attention to the fault tolerance technology in detail. This method helps the collaborative server work in normal and degraded (duple or even single modular) modes, and can tolerate transient or permanent faults. Meanwhile, due to the importance of the server in hybrid P2P server, server application upgrade is impossible without server stop. A novel seamless software upgrade method is brought forward through the intelligent state-transition-control in CS-TMR package to reduce the cost of software upgrade.