A Big Data solution for troubleshooting mobile network performance problems

Big Data has become a major competitive advantage for many organizations. The analytical capabilities made possible by Big Data analytics platforms are a key stepping stone for advancing the business of every organization. This paper illustrates the development of a big data analytics system for mobile telecommunication systems. The authors developed a solution for analyzing data produced by mobile network nodes which contain data relevant for predictive maintenance and troubleshooting purposes. The solution is built around the problem of working with small files in the Hadoop environment. The logs collected from mobile network nodes are small binary files between 5 and 15MB in size. These binary log files need to be decoded to a readable format, and then analyzed to extract useful information. In this paper, the authors provided a benchmark of various scenarios for collecting and decoding the binary log files in a Hadoop cluster. As a result, the scenario with the highest performance has been used in the implementation of our solution. The developed solution has been built and tested on a live Hadoop cluster using real-world data obtained from several telecom operators around the world.