The OpenFlow rules placement problem : a black box approach. (Une approche « boite noire » pour résoudre le problème de placement des règles dans un réseau OpenFlow)

The massive number of connected devices combined with an ever increasing volume of data traffic push network operators to their limit by limiting their profitability. Software-Defined Networking (SDN), which decouples network control logics from forwarding devices, has been proposed to tackle this problem. An important part of the SDN concepts is implemented by the OpenFlow protocol that abstracts network communications as flows and processes them using a prioritized list of matching-actions rules on the network forwarding elements. While the abstraction offered by OpenFlow allows to implement a large panel of applications, it raises the new problem of how to define the rules and where to place them in the network while respecting all technical and administrative requirements, which we refer as the OpenFlow Rules Placement Problem (ORPP). In this thesis, we focus on the ORPP, and propose a black box abstraction that can hide the complexity of rules management. First, we formalize that problem, classify, and discuss existing solutions. We discover that most of the solutions enforce the routing policy when placing rules, which is not memory efficient in some cases. Second, by trading routing for better resource efficiency, we propose OFFICER and aOFFICER, two complementary rules placement frameworks that select and place OpenFlow rules satisfying policies and network constraints, while minimizing overheads. The main idea of OFFICER and aOFFICER is to install rules on efficient paths for important, large flows, and let other flows follow default, less appropriate paths. On one hand, OFFICER is designed based on optimization techniques to solve the offline ORPP, in which the set of flows is assumed known and stable in a period. On the other hand, aOFFICER uses adaptive control techniques to cope with the online ORPP, where the set of flows is unknown and varies over time. These proposals are evaluated and compared extensively to existing solutions in realistic scenarios. Finally, we study a use case of the black box abstraction, in which we target to improve the performance of content delivery services in cellular networks.