Holistic Runtime Performance and Security-aware Monitoring in Public Cloud Environment

Jha, DN; Lenton, G; Asker, J; Blundell, D; Wallom, D

doi:10.1109/CCGrid54584.2022.00128

Holistic Runtime Performance and Security-aware Monitoring in Public Cloud Environment

Lookup NU author(s): Dr Dev Jha ORCiD

Downloads

Accepted version [.pdf]

Licence

This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).

Abstract

© 2022 IEEE. The emergence of cloud computing allows users to execute their applications in a ubiquitous manner. Public cloud offers various ready-to-use services e.g. AWS EC2, Amazon RDS on a pay-per-use basis. Alongside these advantages, the cloud also brings a number of issues, for example offloading data for storage and computation may lead to privacy and security concerns. Also, it is not easy to guarantee the performance of the underlying system. With the increasing performance and security concerns, it is necessary to continuously monitor and evaluate the system and its performance. This can help us to quickly detect anomalies that can hinder system performance and/or make the system untrusted. In this paper, we present PERSECMON: performance and security-aware monitoring framework for continuous run-time monitoring in the public cloud environment. PERSECMON provides not only the system performance metrics but also the security measurements which can be used to analyse the system state at run-time. It uses the BCC/eBPF (BPF Compiler Collection/Extended Berkeley Packet Filters) framework to instrument the system. PERSECMON is integrated with the open-source user interface framework, Kibana which provides a clear visualisation of the obtained metrics. To show the efficacy of our proposed work, we have developed a benchmarking case study using Bonnie++, Fibonacci Sequence and Netperf executed on Ubuntu Server 21.04. The results show that PERSECMON successfully captures relevant metrics that can be utilised in real-time to analyse the system performance. These metrics can further be accessed to detect the system state including memory leaks, queuing delay and remote access time which may lead to security or reliability events.