DynPaC: Dynamic and Adaptive Traffic Engineering for SDNs

The Dynamic Path Computation (DynPaC) framework allows an adaptive and fault tolerant provisioning of connectivity services on top of a SDN/OpenFlow infrastructure. It has been designed to optimize network utilization, while at the same time guaranteeing an appropriate performance of the higher level applications. Therefore, traffic disaggregation is not arbitrarily performed, but taking into account the specific characteristics of the conveyed traffic flows. Additionally, in order to efficiently reallocate traffic flows when a link failure occurs, at the time of computing the optimal path through which a flow is transmitted, a set of backup paths are also computed. The main element of the DynPaC framework is a centralized Path Computation Element (PCE), which computes the best path between a given ingress-egress switch pair according to the demanded bandwidth, real-time information regarding the network topology and the available resources, scheduling information and a set of suitable splitting ratios of the service flows transmitted through the network. Information regarding the different available alternatives to disaggregate traffic flows without degrading applications’ performance is provided by a Traffic Pattern Analyser module. The DynPaC framework has been implemented by extending the Open Daylight Project (ODP) and a first release of an operational DynPaC framework has been demonstrated at the Supercomputing 2014 conference.


Part of session

SDN Applications

Related documents