A carregar...
Publicação
Integrating security and resource efficiency in Microservice Architectures using AI techniques
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 3.22 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
The widespread adoption of cloud-native microservice architectures has enabled highly scalable and flexible distributed systems, but has also introduced significant challenges in observability, resource management, and operational stability. As these systems scale and become more dynamic, traditional monitoring approaches struggle to provide timely and accurate insights into node-level behavior, resource contention, and short-lived performance issues. This work presents the Holistic Secure Framework (HSF), a distributed framework designed to improve observability and coordination in containerized microservice environments. HSF deploys a lightweight agent on each cluster node to collect fine-grained resource-usage and performance data at the source, thereby enabling accurate attribution of resource consumption to individual services. Collected data is securely stored in a persistent data store, with access controlled through a token-based authorization mechanism. The framework introduces two key protocols: CAP-LE, a context-aware leader election protocol that incorporates real-time and historical node metrics into the election process, and a secure heartbeat protocol that continuously verifies agent health and code integrity. These mechanisms support informed leadership selection, strengthen system trust, and enable coordinated responses to failures or abnormal behavior. The proposed framework and protocols are evaluated through experiments on clusters of varying sizes. CAP-LE is compared with Raft and enhanced Raft-based protocols, demonstrating competitive leader-election performance with predictable overhead. Resource usage measurements further demonstrate that the framework’s added security and coordination capabilities incur minimal CPU and memory overhead. Overall, this work presents a practical approach for improving the reliability and resilience of modern cloud-native microservice systems through enhanced observability, resource management, and secure coordination.
Descrição
Tese de Mestrado, Engenharia Informática, 2025, Universidade de Lisboa, Faculdade de Ciências
Palavras-chave
Microservices Architectures Security Containerized Applications Resource Optimization AI-Driven Vulnerability Detection