Authors : Vempaty Prashanthi, Srinivas Kanakala and Subhash Parimalla
Abstract: Modern engineering incorporates clevertechnologies in all factors of our lives. Smart technologies are generating terra bytes of log messages every day to record their status. It is crucial to research these log messages and present usable records (e.g., patterns) to directors, so as to manipulate and reveal those technology. Patterns minimally represent large corporations of log messages and enable the administrators to do further analysis, along with anomaly detection and event prediction. Although, patterns exist typically in automatic log messages, spotting them in large set of log messages from heterogeneous resources without any prior information is a widespread undertaking. We propose a big data using Hadoop that extracts high pleasant styles for a given set of log messages. Our approach is fast, memory efficient, accurate and scalable. Hadoop is implemented in map-reduce framework for disbursed platforms to procedure hundreds of thousands of log messages in seconds. It is a robust approach that works for heterogeneous log messages generated in a wide style of systems. Our technique exploits algorithmic techniques to limit the computational over-head based totally on the truth that log messages are continually routinely generated. We examine the performance of Log-Mine on hugeunits of log messages generated in commercial applications. It has efficiently generated styles which might be as exact as the styles generated by genuine and un-scalable method whilst achieving a 500 per speedup.
Vempaty Prashanthi, Srinivas Kanakala and Subhash Parimalla, 2020. Generating Analytics from Web LOG. Journal of Engineering and Applied Sciences, 15: 3503-3508.