The purpose of the project is to develop performance models for distributed web servers and customer behavior. The models will mainly be used to study the problem of protecting a server from being overloaded when the offered traffic is high. If a server is not protected long queues may be built up in front of the processors and thus the waiting time for customers may get very long.
In traditional telephone networks such problems are solved by using load control mechanisms to protect network elements. If the load becomes high, new calls are simply rejected and are not given dial tone. Such protection mechanisms may also be needed for web servers. Queuing models and simulation will be used to study both stationary and transient behavior of different overload control mechanisms.