Optical networks for cost-efficient and scalable provisioning of big data traffic

Abstract

This article shows how recent advances in optical networks can be used to improve big data processing by cost-effective and scalable provisioning of high-bandwidth connectivity for big data traffic in backbone networks and consequently tackle the current problems related to big data processing in distributed environment including cloud computing. We focus on two optical technologies, namely currently the most popular wavelength division multiplexing and a new emerging approach of elastic optical network (EON). The performance metrics related to both optical network resources (network capital expenditures/operating expenditures (OPEX) cost, power consumption optical spectrum usage) and computing resources (response time and OPEX cost) are investigated in pan-European and US networks based on an example of a big data application related to hyperspectral image processing. Results indicate that provisioning of big data traffic directly in the optical layer with the use of the EON concept provides the best results in most of examined criteria.