Indexed on: 15 Dec '16Published on: 12 Dec '16Published in: Computers & Electrical Engineering
Recent Distributed Denial of Service (DDoS) attacks on cloud services demonstrate new attack effects, including collateral and economic losses. In this work, we show that DDoS mitigation methods may not provide the expected timely mitigation due to the heavy resource outage created by the attacks. We observe an important Operating System (OS) level “internal collateral damage”, in which the other critical services are also affected. We formulate the DDoS mitigation problem as an OS level resource management problem. We argue that providing extra resources to the victim’s server is only helpful if we can ensure the availability of other services. To achieve these goals, we propose a novel resource containment approach to enforce the victim’s resource limits. Our real-time experimental evaluations show that the proposed approach results in reduction in the attack reporting time and victim service downtime by providing isolated and timely resources to ensure availability of other critical services.