Virtual networks prediction by using the super multiplicative DEA model and fractals

Virtual Networks (VN) have been used to support the network traffic in data centres for the delivery of all kinds of services in cloud computing. Here, we developed a super-efficiency multiplicative data envelopment analysis model (SMDEA) for VN service´s forecasting based on real measurements. Another contribution of this essay is to show that the self-similarity (SS) with Long-Range Dependence (LRD) has a different performance per network/setting/device that were analysed as decision-making units (DMU) by the multiplicative DEA models. This paper also employs fractal analysis on computer networks to predict traffic trends using a one-time series evaluation per DMU. Then, the multiplicative DEA models give the decision-maker the capacity to pick a setting with smoother traffic and better TCP transfer rate over time. Finally, the results demonstrate the superiority of SMDEA versus classic super-efficiency DEA models, also providing future research directions.