Take for example a web application running privately in a data center that is taking online orders. When it reaches peak capacity due to high web traffic, an option is to burst and provision servers on a public cloud to handle additional loads so that all orders can be processed on time. Due to the elastic and on demand nature of public cloud computing, this is possible because you can essentially only pay for the time you provisioned servers during the burst event.
Due to its abilities to provide resources on an as-needed basis, cloud bursting is a flexible scaling architecture. It provides cloud consumers with the resources they need, and the foundation is based on automated scaling and resource replication.