Design And Implementation Of The Big Data Analytics Telecom Platform Based On Hadoop

With maturity of the construction of the telecom operator’s network and rapidly development of 3G technology, network quality has become an important topic of concern telecom operators.How to ensure high quality network has become a priority work of the telecom operator’s network optimization. In telecom PCMD bill and the amount of traffic data is on TB level.To meet these requirements will need to analyze large amounts of data on TB level, the original network optimization platform is hard to handle with such a large amount of data. In order to massive data analysis and processing, to quick query and to compile statistics on network conditions accurate, and therefore build a large data analysis platform is imperative for network optimization. Hadoop architecture is recognized as a cost-effective solution to handle massive data in the industry. This solution provides a way to build on ordinary PC servers for high availability, Method of distributed computing and storage of high performance. This solution is particularly suitable for log analysis. Therefore, the implementation of a telecom big data analytics platform based on Hadoop is one of the best choice telecommunications network optimization sector.This paper describes the status of Hadoop technology and introduces Hadoop and related components. This paper introducts especially focusing on HDFS, YARN, Mapreduce, Hive, HBase and working principle of them. It is proved that Hadoop technology for big data analysis has the advantage of telecommunications. For the design and implementation of the big data analytics telecom platform lay a basis for the use of the technologies.Because of the needs of telecommunications such as fast handling customer complaints, analysising cell load, analysising quality of Service, troubleshooting, safeguarding major scene, analysising business accounting, analysising user behavior, analysising mainstream business, this paper introducts system design. With the classification by demand, Hadoop use the characteristics of the different components, and their respective advantages. Quick query feature is achieved by Hbase. Fast statistical analysis feature is achieved by Hive. In addition the middleware technology provides conditions for external system to use of Hadoop resources.This paper describes the design of use Hbase and Hive tables. Found in the use of data skew, etc. the system is optimized and adjusted according to the situation of each task and data features. Finally the system get the performance improvement.Implementation of the big data analytics telecom platform achieve some feature including analysising traffic load, analysising users feeling, real-time monitoring, analysising network quality, analysising security question, complaint handling etc.