Design And Implementation Of Cloud Call Detailed Record System Based On Hadoop

The call detailed record system is one of the bridge between communication service operator and users, user experience of call detailed record system is related to the satisfaction of communication service operator's business. Therefore, low-latency,highly reliable of call detailed record system is the moving target of the operators chased, call detailed record reliable system not only avoided the loss caused by system failure occurred, as well as increased user satisfaction and system reliability. In the cloud computing technology has become more sophisticated today, a variety of cloud technology has been applied to the production and life. Hadoop as the basic framework of cloud computing platform, with its low cost, high expansion, high processing performance to win the majority of user groups.Traditional call detailed record system based approach to the local file system to store the call detailed record data, the amount of data stored in a file is huge, and a higher level of customization file system structure, generally the coupling between actual business is lower, when the query even need to traverse the entire file can find the information what user wants, this causes user to wait too long. Second, in the traditional call detailed record system, data query, data import and other operations are deployed on the same server, causing the server load is too large, stability can not be guaranteed, prone to sudden business interruption and other issues.According to the issues and the communication service operator's actual needs of call detailed record system, this paper presents the design of a cloud call detailed record system based on Hadoop, focus on the user's query and statistical analysis operations, uses the real-time computing framework Storm for dealing with the collected user information, high reliability and low latency, can be a steady stream of processing data information. The data stored in two parts after processing, one is saved in No SQL database HBase, and relying on its properties by column stores can quickly provide feedback to the user query results, and the other saved in HDFS file system,used for the statistical analysis Hive. The system is designed to take advantage of low-cost servers are clustered, with a low cost, high expansion, high processing performance and other advantages, laid a solid foundation for communication service operator in high-traffic background continuously improve the quality of service users.