Research On Multi-federations Auto-deployment Of Grid-based Simulation Platform

Grid based Distributed Simulation Platform (GDSP) is the base infrastructure of Service-Oriented extensible Modeling/Simulation supporting environment architecture. The implementation of GDSP can promote reusability of current simulation supporting environment. Also, it enables a better reuse of existing simulation resources, and obtains more effective management to resource. GDSP provides simulations with fault and load balance capabilities. It also enhances the flexibility and scalability of simulation by organizing the system in loosely coupled style.It is a better method to use bridge-federate to interconnect multiple developed federations in the purpose of building large-scale distributed simulation. Multi-Federations simulation scene Auto-Deploy System (MFADS) is designed to act as a global grid resource scheduler to choose the grid resource automatically for every federate of the multi-federations simulation scene. The MFADS can describe the scene in detail and generate a very detailed description for the Simulation Scene. By analyzing the description of the simulation scene, a performance model is introduced to analyze the network resource requirement of the simulation applications. The China Grid Super Vision (CGSV) can obtain the system information of grid computing resources. Due to the description and the existing grid computing resource, a cluster-based grid resource selection algorithm is designed to automatically choose the resources for every federate of the simulation scene to let the whole simulation run more efficiently. Remote deploy function is also implemented to deploy every federate to the target host chosen by MFADS.Through the system test, this multi-federations simulation scene auto-deployment system can well be applied to large scale distributed simulation system. The test also shows a greatly improvement on the function of HLA in the sharing of resources and associated coordination; enables dynamic allocation of resources in grid simulation and greatly enhances the efficiency of the system.