Research On Search And Location Mechanism Based On Digital Organism Database System

Along with the development of IPTV(Internet Protocol Television), EC(Electronic Commerce), Web Service, download of music and movies as well as other large scale network applications, more and more distributed computing technologies are introduced, such as Grid, P2P and cloud computing which has been popularly used nowadays. To settle particular problems in distributed network environment, these technologies generally employ existing foundational software platform, leave alone some constraints in traditional distributed system and develop new technologies or improve existing technologies. So, no one technology among them can do for others up to now.To achieve all-purpose computing pattern and integrated solution, 8010 research group lead by Prof. Xinsong Liu introduces large scale distributed technologies to foundational software layer and has successfully developed a platform named Digital Organism System (DOS) by exploring with many years and many phases. DOS mainly includes two kernel software—Digital Organism Operating System (DOOS) and Digital Organism Database System (DODBS). Today, DOS has been applied to practical applications and affirmed by many fields. Besides has all functions just like other foundation software, DOS must also face to the high performances problem about network service like other large scale distributed application systems. Like all other large scale application software, solution of this problem in DOS mainly relies on the search and location mechanisms about sharing resources in its own system.This dissertation deeply and widely studies on searching and locating mechanism of DOS based on DODBS. The main research contents and major contributions are as below:1. Based on the analysis about search and location mechanisms, research status and development trends of large scale application systems based on Grid, P2P and cloud computing as well as other distributed computing, the derivation, goals and significance of DOS is introduced, the three phases developing history of DODBS is summarized at the first time and the network construction, system architecture and applications of finished product in each phase are introduced in detail.2. The hybrid overlay network of DODBS, which is composed of three layer networks including the highest layer of Districts, the middle layer of Sites and the lowest layer of Nodes and peer-to-peer network in highest layer, is constructed. Based on the logic architecture, the search and location model is established, which includes distributed and parallel pattern in Site levels, backtracking pattern in Area levels and peer-to-peer pattern in highest layer. The goals, classes, processes and characteristics of search and location based on DODBS are introduced.3. Many special datasheet, which can really assist improve performances, is put forward and organized. Such as the cache information table witch can be used to make use of hot resource caches, the information-database witch logs search and location histories, the repository witch describes related search knowledges and the sharing resource information based on names and keywords witch describes the sharing resources themselves. At every end of one datasheet introduction, the corresponding search and location algorithm is described.4. The searching and locating mechanism based on distributed and parallel technology in Site level is studied. After the introduction of distributed and parallel transactions and the classification of server roles in the processes of transaction, the distributed and parallel 2PC based on the consistency of data directory is described in detail. Finally, some limitations of distributed and parallel searching and locating pattern illustrate it is not suit to extend to Wide Area Network.5. A special of information compression method based on location index of keywords is present. First, the theory and process with bit-set is described in detail. Next, the algorithms of index publishing, integrating and searching in District based on this compression method is introduced and analyzed. Finally, the validity of this compression method and the efficient performances of related algorithms are analyzed and validated.6. A kind of replica location method of MQoSRL which is based on multi-dimensional QoS constraints is supposed. First of all, the problem of multi-dimensional QoS constraints is defined and forth described in formalization. Then, the whole algorithm based on MQoSRL within system domain is introduced and analyzed. Finally, the higher efficiency, validity and load balance as well as the capability to put down the system average response time of MQoSRL are illustrated by analysis and application results.