Design And Implementation Of Big Data Based BGP Routing Events Analysis System

Large-scale BGP routing change events can cause connection interruption in network application service, reduction of available bandwidth and traffic jam in link entrance, which affect network and service performance seriously. Traditional methods for identifying BGP events are based on BGP RIB or update messages. For those methods based on update messages, some use few monitors which only provides limited view of the Internet toplogy and can not monitor global Internet. But others which analyze aggregated information from each monitors (count update number or type) may overestimate the impact of monitor-local events.Some works analyze BGP RIB messages to identify disruptive incidents,but they are limited by the collection time interval of popular routing information collection project, which lead to coarse time granularity.The thesis proposed a fine granularity method to indentify BGP events.The method combines RIB and update messages of BGP to build a new routing change matrix with small granularity. It uses decomposition method to identify large-scale BGP routing events in short time interval and adds constraint to avoid local prefix events. On the basis of the proposed method, the paper designs the BGP routing events analysis system based on Spark. It includes data layer, the business logic layer and application layer. The data layer collects information from a large number of monitoring points for the system as the data source. The business logic layer formats the origin data to satisfy upper layer demand, and the application layer realizes detection work of indenfying large-scale routing event from such data. Finally, the system is tested to prove that the method can identify large scale BGP events in short time and the system satisfies functional and non-functional requirments.