Quality Control: A Key to Stress-Resilient Internet Services

Today´s Internet services are largely vulnerable to flash crowds-like DDoS attacks because the visits in such attacks often bear much resemblance to normal accesses and most of them cannot be detected with certainty. Current same-service-to-all models can hardly protect the interests of legitimate clients from the interference of suspicious traffic. Application-level quality control in edge servers relies on autonomic resource management technologies to offer different levels of service quality to different classes of traffic according to server load conditions. By downgrading the performance of aggressive visits, the model ensures resource fair sharing and builds a firewall around legitimate clients. Because the quality control model can isolate the performance of different traffic in stressed servers, it also offers a strong incentive for differentiated pricing policies for the next generation of Internet services. Challenge issues with the quality control model will be discussed. An autonomic quality control framework for web servers will be presented.