Search

The infrastructures of SANs are complex, and the implementations/designs of SANs are varied. So finding a proper modeling method to analysis and simulate SAN performance in a systemic/consistent way is very important for the work efficiency of SAN designers and performance tuners. This paper proposes a methodology of using a multi-tier multiserver queue (MMQ) network system to model a SAN system, which can be generalized to different SAN instances neatly, and be used for both performance analysis and simulation. Besides the main idea and high-level MMQ model, the application of this method is illustrated by constructing a software simulator, which is also designed and implemented in a systemic and reusable style. The simulation validation shows that maximum error range is less than 10%. So we can consider this method a sound solution for the quick performance design comparison and analysis of SAN.

IMDSS is an intelligent database system for market analysis and sales forecast. One of its novel features is its output of analysis and forecast report documents presented in Chinese speech with graphs, data tables and 3D complex model real time display. The contents and organization of this report document as well as its multimedia presentation are completely specified by a script language called Moulan. This paper introduces the structure of Moulan, and explains its usage in specifying what to be included in the report and how multimedia presentation should be arranged. A Moulan script file is processed by IMDSS in two steps-its translation and execution. In the translation step, a series of internal files for the execution of the script are generated. In the execution step, these files are used to create the multimedia output and synchronize the output of different media. The processing of Moulan script is discussed in detail in the paper.

Mars is both a spectrum and a structured knowledge based multiple agent interpretation system for remote sensing images. It represents the required knowledge in various schema closest to its original form by a blackboard structure. Mars uses a working-frame to hold a description of the traits and hypotheses addressed by agents. It supports a set of standard operations used to manipulate the data in the blackboard by all agents to integrate heterogeneous knowledge representation. By using a combined method of dynamic trait extraction and level conflict resolution respectively, Mars handles conflicts within and among agents. This paper discusses the design and implementation of Mars, including its architecture, mechanisms and implementation environment, and future work.