Anomaly detection on Kubernetes ecosystem: an AI and data-driven approach

Bozorgi, Fatemeh (2024) Anomaly detection on Kubernetes ecosystem: an AI and data-driven approach. [Laurea magistrale], Università di Bologna, Corso di Studio in Artificial intelligence [LM-DM270], Documento full-text non disponibile
Il full-text non è disponibile per scelta dell'autore. (Contatta l'autore)

Abstract

In today’s digital landscape, computing is essential for innovation and efficiency across sectors. Cloud computing has revolutionized this space by offering scalable, cost-effective, and globally accessible services, reducing IT costs and enabling rapid growth through automatic updates. The trend towards larger data centers is driven by economies of scale, virtualization, and the rise of web-based applications. However, the complexity of cloud systems makes them prone to runtime issues like hardware failures and software faults. Ensuring faultless operation and optimizing resource use is crucial, as downtime can lead to significant financial losses. Anomalies pose challenges to cloud reliability, causing performance degradation and potential system outages. Autonomic anomaly detection has become critical, enabling cloud systems to self-manage by detecting and responding to abnormal behaviors to maintain system reliability. Effective monitoring and performance data collection are key to identifying anomalies, and ensuring scalable, flexible cloud services remain reliable. Kubernetes, an open-source container orchestration platform, plays a key role in managing cloud environments by automating the deployment, scaling, and management of containerized applications. It improves resource utilization, scalability, and system reliability, making it essential for modern cloud operations. This study explores the integration of Kubernetes for anomaly detection in cloud computing to enhance reliability. Using a web-based application as a case study, various techniques were evaluated, including statistical methods, machine learning models, and deep learning approaches. Deep learning approaches achieved the highest anomaly detection accuracy (0.982 F1 score). Integrating this system within Kubernetes significantly improved uptime, reduced error rates, and sped up anomaly responses, demonstrating Kubernetes’ potential to enhance cloud system efficiency and minimize downtime.

Abstract
Tipologia del documento
Tesi di laurea (Laurea magistrale)
Autore della tesi
Bozorgi, Fatemeh
Relatore della tesi
Correlatore della tesi
Scuola
Corso di studio
Ordinamento Cds
DM270
Parole chiave
Anomaly Detection, Machine Learning, Cloud-Native Systems, Time-Series Analysis, Kubernetes, AI-Driven Analytics
Data di discussione della Tesi
8 Ottobre 2024
URI

Altri metadati

Gestione del documento: Visualizza il documento

^