Contents

Web sites are constantly threatened by sudden surges in traffic, also known as flash crowds. Current systems deal with flash crowds by utilizing replication or edge computing to dis-tribute the load between multiple servers. However, when dealing with dynamic websites, the database part of the system quickly becomes the bottleneck. As a result, current systems are unsuitable when applied to dynamic websites. In this thesis, we introduce flashCBC, a content-blind query result caching middleware designed to unburden the database server during a flash crowd