Arquitectura de Gestión Centralizada de Logs con OpenSearch e IA en los Sistemas Institucionales de la Universidad Nacional de Loja

This project addresses the issue of decentralized log management in the institutional systems of the Universidad Nacional de Loja (UNL), which hinders early detection and resolution of incidents, negatively impacting the security, operability, and continuity of technological services. To tackle this, the present Graduation Project (TT) aims to design a centralized architecture for log management in UNL's systems using open-source tools such as OpenSearch and artificial intelligence (AI). To achieve this objective, the methodology integrated modern technologies like Fluent Bit, Apache Kafka, Redis, OpenSearch, and Large Language Models (LLMs) such as LLaMA3, Mistral, and Gemma2, designed to collect, process, and analyze large volumes of data. The test cases, conducted in controlled environments, demonstrated the system's ability to centralize data, identify anomalies, and provide recommendations for decision-making. The impact of log centralization was measured based on incident response times, showing a significant improvement compared to manual processes. Specifically, incident response and resolution times were shorter with the proposed architecture. The implementation of this centralized architecture, based on open-source tools, not only improves operational efficiency and incident response capabilities but also provides a sustainable and accessible framework for educational environments. This project contributes to the academic community by demonstrating the feasibility of integrating open-source technologies and artificial intelligence for log management, offering a reference framework for other educational institutions seeking to enhance their monitoring and incident response capabilities in their technological systems.

• 第一著者: Macas Loja, Edgar Manuel (author)
• フォーマット: masterThesis
• 言語: spa
• 出版事項: 2025
• 主題: GESTION CENTRALIZADA DE LOGS; ARQUITECTURA CENTRALIZADA; ANALISIS DE LOGS
• オンライン･アクセス: https://dspace.unl.edu.ec/jspui/handle/123456789/32133