Cybersecurity in High-Performance Computing: Safeguarding Data and Architectures

Abstract

High-Performance Computing (HPC) systems are essential in fields such as climate modelling, drug development, genomics, artificial intelligence, and national defence. Due to their rapid expansion in scale and connectivity, these systems have heightened exposure to advanced cyber threats that threaten data integrity, confidentiality, and system availability Conventional cybersecurity measures, tailored for standard IT systems, sometimes fall short in HPC environments because of the distinct architectural and performance requirements of parallel computing and exascale architectures. Key challenges include the lack of a secure perimeter, the complexity of HPC systems, and the increasing use of cloud-based HPC and AI, which necessitate a shift to more dynamic and performance-aware security solutions to safeguard sensitive data and maintain system integrity.

This study examines the changing threat landscape in HPC systems, concentrating on vulnerabilities at the data, network, and architectural tiers. It offers a comparative examination of traditional IT and HPC security requirements, investigates significant cyber events using HPC clusters, and assesses performance-security trade-offs. The study provides a multi-layered defence system that integrates encryption, access control, intrusion detection, and policy compliance techniques specifically designed for HPC workloads This study demonstrates how the integration of technical safeguards and governance models enables HPC to maintain resilience against cyberattacks while preserving its essential function in scientific and industrial innovation..