Institutional Repository
| 广域网服务发现机制的研究 | |
| Alternative Title | Research on Service Discovery Mechanism for Wide Area Network Environment |
| 张英 | |
| 2007-06-03 | |
| Degree Grantor | 中国科学院软件研究所 |
| Degree Level | 博士 |
| Place of Degree Grantor | 软件研究所 |
| Keyword | 服务发现 服务描述模型 索引后缀树 Qos感知的协作发现 语义匹配 |
| English Abstract | 随着Internet和分布式系统的广泛应用和发展，人们可以在世界范围内共享各种服务和信息。网络中的服务纷繁复杂，种类众多。因此，有效地发现、管理各种服务信息至关重要。 服务发现是网络分布式环境下进行信息共享、数据集成、流程协作的前提。网络用户可以在互连的网络环境中发布可共享的服务信息，或者查询自己需要的服务资源，以便进行进一步的信息处理。目前学术界和工业界在服务发现方面进行了不少的探索和研究，但是针对广域网的服务发现还存在较多问题有待解决。例如，在广域网环境中，还没有可被服务发现机制直接使用的广播和组播机制，同时，设备和服务的种类和数量都非常庞大，服务发现必须具备规模可伸缩性，等等，这些问题在已有的服务发现系统中尚未得到有效地解决。 本文主要对广域网环境下的服务发现机制进行研究。根据广域网服务具有多样性和异构性的特点，本文提出了一种支持多样性、屏蔽异构性的服务描述模型SDMDH，并基于该模型，构建了具有一定可伸缩性的服务发现系统OSCAR（Once Service CAtalognet for RFID environment）。OSCAR采用由目录服务器组成的应用层覆盖网络来分布存储系统中的服务信息，同时，采用基于软状态的服务信息管理维护策略，以适应服务的动态变化。系统基于后缀树形式的服务信息索引，提供了多种服务发现机制，其中包括QoS感知的协作式服务发现，即根据客户的QoS指标生成应用层选播/多播路由，并基于此实现协作式服务发现。目前，系统不仅提供了精确匹配与部分匹配策略，还支持语义匹配，以提高服务发现的智能性。系统还基于历史查询记录将“热点”服务向查询密集的目录服务器复制以实现负载平衡。 |
| Abstract | Along with the development and extensive utilization of Internet and distributed systems, people can share services and information around the world. Since services in the network are complex and diversified, the effective discovery and management of service information is essential. Service discovery is the premise of the information sharing, data integration and process collaboration in the distributed environment. In interconnection network environment, users can publish the shared services’ information, or inquire about their needed resources for further information processing. In recent years, some research work from both academy and industry has been done for solving this problem. But after investigating their work, we find that the work aiming at services discovery for wide area network is still deficient. In the wide area network environment, there is no suitable broadcast and multicast mechanism available to service discovery directly, meanwhile, the kind and quantity of equipments and services are enormous, so scalability with respect to size of services and concurrent users should be achieved, and so on. The above issues have not yet been addressed effectively in the existing services discovery system. This paper focuses on service discovery mechanism for the wide area network environment. Keeping in mind the diversity and heterogeneity of services, we propose a service description model SDMDH which can describe diverse services and hide their heterogeneity, and then we construct a service discovery system OSCAR with high scalability. OSCAR is built on an application overlay network which consists of directory servers and is used to store service information in a distributed way. Soft-state based service management scheme is adopted by OSCAR to better adapt to the service’s dynamic changes. Based on the index of suffix tree, OSCAR provides multiple service discovery policies, which includes QoS-aware collaborative discovery, i.e., generating anycast/multicast routing path first according to client’s QoS metrics, then achieving service discovery. OSCAR provides not only accurate matching and partial matching, but also the support to semantic matching to improve the intelligence of service discovery. Based on historical query records, OSCAR can replicate the "hot" services to the query-intensive directory servers as a scheme to support load balancing. |
| Pages | 84 |
| Language | 中文 |
| Content Type | 学位论文 |
| URI | http://ir.iscas.ac.cn/handle/311060/6388 |
| Collection | 中科院软件所_中科院软件所 |
| Recommended Citation GB/T 7714 | 张英. 广域网服务发现机制的研究[D]. 软件研究所. 中国科学院软件研究所,2007. |
| Files in This Item: | ||||||
| File Name/Size | DocType | Version | Access | License | ||
| 10001_20042801502901（1191KB） | 限制开放 | -- | Application Full Text | |||
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.
Edit Comment