【作者】 李杰；

【导师】 王国仁；

【作者基本信息】 东北大学 ， 计算机技术（专业学位）， 2013， 硕士

【摘要】 随着云计算研究的兴起,云中计算资源和存储资源的优点不断被发掘,主要包括：分布式、大规模、动态性等。因此,越来越多的应用系统开始向云平台进行迁移。然而云平台与传统单或多服务器模式的区别也给系统开发带来了一定挑战。本文依托于国家海洋公益项目,研究并实现了基于HBase的海洋信息查询系统,使用户可通过图形接口检索海风、卫星图片等信息。通过对查询技术的有效应用,系统可以高效地从海量数据信息中提取用户关心的海洋基本数据信息,便于利用该信息为渔业生产和海洋灾害预防进行服务。基于HBase的海洋信息查询系统框架分为三个层次：数据存储层、数据处理层和操作与显示层。数据存储层采用HBase与HDFS进行实现,其中利用HBase对文本信息进行存储,HDFS对图片信息进行存储；数据处理层是本文的实现重点,该层实现了海洋信息数据的上传与下载功能,针对用户输入的条件对数据进行查询；操作与显示层则采用Java Swing实现了系统与用户的交互窗口,用户可通过该窗口输入查询条件,观察查询结果,对结果进行进一步操作。在海洋信息查询系统实现过程中,本文对查询优化技术进行了研究与改进。首先在NearGoos气象数据查询过程中,采用B-order值的算法对二维空间数据进行划分,结合HBase数据表中rowkey的排序特点,对空间查询进行优化,使查询效率有了两个数量级的提升；而在实现MODIS空间查询时,本文采用了射线法用于判断空间查询条件与MODIS图像的位置关系,并对其进行实现；最后在操作显示层实现时,采用生成缩略图的方式解决Java显示图片占用内存过大的问题。本文首先结合查询系统的运行示例对查询系统进行展示,然后对系统中使用的HBase查询优化技术进行测试与分析,验证了其良好的可扩展性和可用性,最后对海洋信息查询系统进行性能测试,结果表明其在云环境中具有高可扩展性与高可靠性。更多还原

【Abstract】 With the rise of cloud computing research, the advantages of cloud computing and storage resources are discovered continuously:distributed computing, massive, dynamic and so on. Consequently, more and more application systems begin to migrate to the cloud platform. However, the difference between cloud platform and traditional single or multiple sever model bring certain challenges to the system development. In this thesis, a marine information query system based on HBase was studied and implemented rellied on National Marine Public Welfare Project. The query system can make users to retrieve the information of sea wind and satellite images by the graphical interface. Through the effective application of query technology, the system can efficiently extract the concerned marine data information for the fishery production and marine disaster prevention.The marine information query system based on HBase can be divided into three layers: data storage layer, data process layer and operation display layer. Data storage layer was implemented based on HBase and HDFS:HBase was used to store the text information and HDFS was used to store the image information. Data process, as the key point of this thesis, was implemented the functions of upload and download of marine information, it can query marine information rely on the limited conditions. Operation display layer implemented the interactive window between system and users based on Java Swing, users can input query conditions, observe result and make further operation.In the implement process of marine information query system, this thesis studied and optimized the query technology. Firstly, in the query process of NearGoos data, B-order was used to divide the two-dimensional space and combine the sort features of rowkey in the table of HBase to optimize the spatical query. The algorithm made the query efficiency increase two orders of magnitude than before. In terms of the spatical query of MODIS data, this thesis used and implemented the ray method to identify the relationship between spatical query condition and the location of MODIS data. Finally, operation display layer used the method of generating thumbnails to solve the Jave display image problem of occupying large memory.This thesis firstly displayed the marine information query system with the running examples, and secondly tested and analized the HBase query optimization technology to vertify its great expandability and usability. Finally, the performance test of marine information query system indicated that the system had high scalability and high reliability in the cloud environment.更多还原