Abstract:

The widespread adoption of Cloud Computing is largely attributed to its on-demand access, self-service capabilities, and pay-per-use model. However, this model also introduces new vulnerabilities—particularly to Denial of Service (DoS) attacks—which not only degrade service quality but also increase resource usage costs. The longer it takes to detect such attacks, the higher the financial burden. Stealthy DoS attacks are especially concerning, as they are designed to avoid detection while still being as disruptive as brute-force attacks. These attacks exploit the system’s worst-case performance using carefully crafted traffic patterns—periodic, pulsing, or low-rate. In this paper, we propose a strategy to engineer stealthy DoS attack patterns that gradually increase in intensity. These patterns aim to maximize financial impact on cloud customers, all while staying within the limits enforced by detection systems regarding job size and service request rate. We detail the implementation of this strategy and analyze its impact on a cloud-based target system.

Description:

.