Skip to Main Content
This paper discusses an attack on the cloud computing model by which an attacker subtly exploits a fundamental vulnerability of current utility compute models over a sustained period of time. Internet-accessible cloud services expose resources that are metered for billing purposes. These resources are subject to fraudulent resource consumption that is intended to run up the operating expenses for public cloud service customers. The details and significance of this attack are discussed as well as two detection methodologies and there respective experimental results. This work investigates a potentially significant vulnerability of the cloud computing model that could be exploited from any Internet connected host. Well-crafted transactions that only differ in intent but not in content are challenging to differentiate and thus this attack may be difficult to detect and prevent.