Operator-Friendly Traffic Engineering in IP/MPLS Core Networks

摘要

Global Internet traffic has increased tremendously and this trend is anticipated to continue for the next years. Consequently, network's performance and users' satisfaction cannot be guaranteed. The necessity for adequate bandwidth to carry this increased traffic has led to the addition of resources to the current core network and service infrastructures, thus affecting the levels of the consumed energy and generally resulting in higher OPerational EXpenditures (OPEX). Obviously, tackling such growth requires sophisticated Traffic Engineering (TE) and associated management schemes. On the one hand, TE mechanisms should be intelligent and self-adaptive so that to take fast and reliable decisions with respect to traffic allocation into network paths. On the other hand, the management of this intelligence cannot rely on the traditional command and control paradigm. Contrarily, it needs to be based on systems that hide technology complexity from the operator and relax him from the rather slow and error prone task of manual configuration. Accordingly, in this work, we present an operator-friendly management framework that is used to drive the decisions of an autonomous algorithm for TE in IP/MPLS core networks. Through the framework, the operator is able to select from a set of high level policies, which the proposed TE algorithm needs to take into account while seeking for routing configurations during its autonomous operation. The behavior of the proposed TE algorithm under the operator choices is experimented through numerous simulations and extensive test cases. Results showcase the efficiency and optimal performance of the algorithm, compared to other TE solutions proposed in literature, while at the same time they validate the framework's friendliness towards operator.