Abstract Cloud technology is becoming a prominent tool for delivering web applications. However, research is scarce on the impact of Distributed Denial of Service (DDOS) on clouds and how various attack methods compare to robust cloud designs. Can cloud technology withstand these attacks, or will traditional web architectures, enhanced with efficient caching and reverse proxy solutions, consistently outperform? This study addresses... Cloud technology is becoming a prominent tool for delivering web applications. However, research is scarce on the impact of Distributed Denial of Service (DDOS) on clouds and how various attack methods compare to robust cloud designs. Can cloud technology withstand these attacks, or will traditional web architectures, enhanced with efficient caching and reverse proxy solutions, consistently outperform? This study addresses these questions by evaluating two architectural models: SlapOS and a setup featuring Nginx at the front and several web servers behind it, under various attack scenarios such as DDOS, Slowloris, and RA Flood attacks. Read More Read Less

Abstract Identifying irregularities in command shell activities is crucial to maintaining computer security. Significant progress has been made in deep learning and natural language understanding, especially with transformer-based models, which have shown considerable potential in tackling intricate security issues. In this study, we present a thorough strategy for spotting irregularities in Unix shell activities using a pre-trained DistilBERT model, combining both... Identifying irregularities in command shell activities is crucial to maintaining computer security. Significant progress has been made in deep learning and natural language understanding, especially with transformer-based models, which have shown considerable potential in tackling intricate security issues. In this study, we present a thorough strategy for spotting irregularities in Unix shell activities using a pre-trained DistilBERT model, combining both unsupervised and supervised learning methods to spot unusual activities while reducing the need for manual labeling of data. The unsupervised technique focuses on understanding Unix shell commands basic structure and grammar, allowing for the recognition of deviations from typical behavior. Trials on a large-scale dataset from real-world systems have proven the effectiveness of our strategy in identifying unusual activities in Unix shell sessions. This research underscores the possibility of using recent breakthroughs in transformer technology to tackle significant security issues in computing. Read More Read Less