Anomaly Detection and Root Cause Localization in Virtual Network Functions

Abstract

HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Experience Report: Log Mining Using Natural Language Processing and Application to Anomaly Detection

“…In our previous work [19] we analyzed monitoring counters such as CPU consumption or number of disk accesses for anomaly detection. Results from counter-based detection showed a good predictive performance that is yet not fully aligned with the results of this study.…”

“…We emulate errors by means of injection tools that implement systems stressing. These tools were used in our previous work [19].…”

“…It has been used for NLP in some previous works but not for the analysis of logfiles. In comparison, our previous work [19] focuses on anomaly detection based on monitoring data collected by means of a specific software agent, deployed beforehand on target machines, and providing numerical metrics on the system behavior. Here we exploit the default system-produced textual logs to predict stress.…”

LUMEN: A global fault management framework for network virtualization environments

“…Recent efforts [19,32,27,29,30], introduced automation in the fault management steps. For instance, Hounkonnou et al [19] proposed a self-modeling concept to face the problem of scalability of the IP Multimedia Sub-system (IMS).…”

“…The automation of the steps was also introduced in some recent research papers that address SDN and NFV environments with self-diagnosis methods using Bayesian networks for certain efforts [27]. While others [29,30], propose Self-healing for NFV chain failures. Figure 2 summarizes the different fault management steps and presents a new fault management outline with the Self-X concepts.…”