Autonomic overload control

The problem

Quality of Service (QoS) management for web-based applications is typically considered a problem of system sizing: sufficient resources are to be provisioned to meet quality of service requirements under a wide range of operating conditions. While this approach is beneficial in making the site performance satisfactory in the most common working situations, it still leaves the site unable to face sudden and unexpected surges in traffic. In these situations, in fact, it is impossible to predict the intensity of the overload. The architecture in use, although over-dimensioned, may not be sufficient to meet the desired QoS under every possible traffic scenario. For this reason, unexpected increases in requests and, above all, flash crowds are considered the bane of every internet based application. Our research is motivated by the need to formulate a fast reactive and autonomous approach to admission control, able to guarantee the desired level of QoS even under unexpected and sudden events as flashcrowds.

Self-* verload control policy

We contribute an original Self-* Overload Control (SOC) algorithm that self-configures a dynamic constraint on the rate of incoming new sessions in order to guarantee the fulfillment of the quality requirements specified in a Service Level Agreement (SLA). SOC enables some fundamental self-* properties such as self-configuration, self-optimization, self-protection.

The algorithm is based on a measurement activity that makes the system capable of self-learning and self-configuring even in the case of rapidly changing traffic scenarios, dynamic resource provisioning or server faults. It does not require any prior information about the incoming traffic, or any manual configuration of key parameters.

Resources available on this website

From this web site you can download the simulator we have used in our papers as well as a detailed manual containing the istructions to set up and execute simulations.

We hope this resources can be useful to ensure a timely dissemination as well as the possibility to compare our solution with others. For more information do not hesitate to contact the authors.

Valid XHTML 1.0 Strict | Valid CSS | OpenSource | Made with a Mac