The Design And Implementation Of Cloud Storage System For Meteorological Information Based On Hadoop

Meteorological information storage system is mainly used in the provinces and cities or national meteorological data storage, display, management and retrieval, is the basis of the meteorological forecast work department, and is also used in the meteorological department one of high frequency information system. At present, as the major department meteorological information amount of data and meteorological data type continuous high-speed growth, how to efficient and reliable storage, management, use of the rich became important meteorological data information, traditional way of meteorological data storage file is stored or distributed relational database storage, the storage solution is increasingly unable to meet the requirements of users to fine weather forecast business.This thesis designed a based on Linux OS and meteorological information of HDFS cloud storage system can be deployed on the server, is easy to enlarge the processing capacity of the storage system, the reliability of the data at the same time also saved the cost for the meteorological department. Meteorological information cloud storage system composed of distributed file systems, local area communication network, server cluster system, virtualized storage technology, such as the distribution in all kinds of storage devices on the network into a resource pool, as a service system provides a unified access interface. This not only can effectively integrate storage resources, but also can improve the convenience of storage and resource sharing. Users to upload meteorological files are stored in a mass data storage pool; When a user need to find and download resources, can freely and easily downloaded from the storage pool, high flexibility. User layer with good scalability, clear architecture of SSH(Spring + struts 2 + Hibernate) framework implements a Web access interface, users and administrators in the case of don’t need to understand the underlying storage details, also can carry on the visual management to the meteorological data, retrieve, and maintenance.In this thesis, finally the development of a complete system advantages as follows: one, to be able to store huge amounts of data. Concurrent access system in more than one storage equipment, work together, not only increase the storage capacity is also raised the speed of the storage. Second, the system can free extension. Can access the high frequency data to copy to strategy, in does not affect the whole system operating under the dynamic capacity expansion. Three, data storage transparent. Back-end data storage space is automatically assigned by the Hadoop system; users do not need to consider the data storage and storage location. Fourth, the reliability of the system is high. Hadoop using fast recovery algorithm, multiple mechanisms and a copy of the calibration function to ensure data integrity and reliability of data users don’t need to care about how to preserve, or adopt what kind of format, data storage will go wrong. Five, the cost is low. Adopting open source free Hadoop framework and common X 86 servers can save a lot of meteorological department funds, ease the tight funding of provinces and cities. So this system can provide large amounts of meteorological information storage with a complete set of solutions, the meteorological department will have broad prospect of application.