A Generic Learning Approach to Modelling Netflows from Historic Observations

Abstract

In this paper we present a generic learning algo- rithm that models the communication patterns between services. Current service landscapes especially in federated environments are characterized by a huge number of services and by a high de- gree of change. In this paper we present a method for quantifying the communication patterns between services in a autonomous fashion to allow predictions of future usage patterns in the service landscape for optimization and simulation. The proposed learning algorithm uses machine learning techniques and generates a probabilistic model based on observed network flow information. We perform the learning and evaluate the learning algorithm based on real world netflow data captured on a cloud testbed. The paper finally discusses potential applications of the proposed algorithm in a autonomous optimization framework for service management.