Network Flow Data in the Cloud

Everything old is new again. Or rather, everything you thought was old is as relevant to today’s distributed service-oriented architecture as it was in the days of manual OSPF metric tuning. Traditional network engineering techniques are based on discrete math – namely, graph theory. A network graph provides a visual and quantitative foundation for analyzing network behaviors to optimize data flow, routing, and resilience in complex topologies. Huge benefits await those able to apply these lost arts to large-scale cloud infrastructure. In this talk, we’ll review those traditional methods, then apply them. We’ll explore how to build network traffic attribution on a per-service level — all without spending piles of money on vendor logging solutions.