Impact of Metro-Embedded Data Centers on Metropolitan Network Design and Traffic Profile

Abstract

Traditional data centers (DCs) that are located in distant areas usually lack flexibility and also have long service latency for metro users. On the other hand, the recently introduced and implemented metro-embedded data center (ME-DC) architecture, based on micro-data centers (mDCs), delivers more flexible services with less access latency. Hierarchical software-defined network control is employed to virtualize and holistically coordinate the distributed mDCs and metro network slices into a reconfigurable virtual data center. Such an approach improves the overall resource mobility and access to the service; however, the impact on the metro network traffic profile induced by user/mDC and inter-mDC connections due to the dispersion of mDCs has not been evaluated. In this paper, we develop a mixed integer linear programming model that optimizes the total DC traffic, taking into consideration the intra-, inter-, and external DC communications. Furthermore, we numerically assess the changes in the metro network traffic induced by ME-DC architecture, considering different types of DCs and their corresponding traffic profile. Simulation shows that the distribution of the mDCs has a beneficial effect, leading to a fourfold reduction in the total traffic load.