Formal Model Of Grid Component And Optimizing Algorithm

Component based software engineering is the necessary tendency of software technology in the next century. It is also one of the hottest topics in current software engineering realm. As traditional component technology is integrated into Internet further, and software architecture based on Grid, grid component software architecture which bases on heterogeneous and distributed network environment and grid architecture, we called Grid-Component, has already become one of the most important directions of component development. The characteristics of Internet environment, such as instability, autonomy and heterogeneity, etc lead the running environments of Grid-Component to an unpredictable situation. Also, the continuous improvements of the network service module, the complexity of Grid-Component presents exponential endless upgrades and increases. So the stability and efficiency of software architecture based on Grid-Component will be confronted with great challenges.This article mainly includes:1. As the new features of software component did in grid architecture , the grid component is proposed. Also the formal definition of software architecture based on grid component is described in detail.2. With the analysis of efficiency influence which made by workflow and dataflow to remote invoking, the formal model of concurrent invoking algorithm(CCCA) and reconstruction dataflow algorithm(DFDM) is discussed. At last, the experiment results show that the system efficiency has great improved by two algorithms.3. Through the basic logic of remote invoking, the connection process which is the most inefficiency step of remote call is found out. The efficient buffer model has the forecast algorithm for remote invoking, including IF(Invoke-Frequency)algorithm, SB(Sort-Based)algorithm and HA(History-Analysis)algorithm. The forecasting arithmetic is discussed and analyze in different areas.4. With the self-healing theory, the software architecture based on grid component is modeled as a biologic self-healing model. The formal definition and function is defined as sensor logic, DNA logic, heart logic, evolve logic, replication logic. The system running performance is tested under evolving algorithm and variation algorithm.5. With the reconstruction the invoking logic theory, the nesting invoking algorithm, recursive invoking algorithm and flow-based invoking algorithm is discussed. Comparing with traditional serial-invoking algorithms, we analyze which algorithm is the best under certain circumstances.6. Finally, the relation of service resources in grid system is proposed which include produce relation, consume relation, support relation and restrict relation. After given the definition of network influence gene, the balance of network resources and the deploy method for optimizing algorithm is discussed.