Bandwidth Reservation for Tenants in Reconfigurable Optical OFDM Datacenter Networks

Abstract

The optical datacenter networks need periodical reconfiguration in response to traffic change. Before the networks reconfigure, the tenant requests are given and should be served within fixed transfer time and spectrum capacity. In this paper, the planning problem of serving the requests in optical orthogonal frequency division multiplexing datacenter networks is investigated. We introduce the knapsack-based spectrum and time allocation (KSTA) problem. The objective of this paper is to maximize the network throughput. We formulate the KSTA problem as an integer linear programming (ILP) model. However, ILP cannot find the optimal solution for large input requests within shorter time. To solve the problem, three fast heuristic algorithms, i.e., the most spectrum first, the most time first, and the most data volume first, are proposed to achieve suboptimal solutions. Furthermore, the simulated annealing (SA) algorithm is employed to yield a better suboptimal solution. The simulation results indicate that ILP provides an optimal solution for small input requests, whereas the three heuristic algorithms and SA can yield suboptimal solutions for large input requests. The results also show that the suboptimal solution to SA is better than those provided by the three heuristic algorithms.

HFFC: A High-Fan-Out and Flexible-Cluster Optical Network Architecture for Data Centers

“…Thus, a traffic demand measurement is employed in this paper. That is, in a reconfiguration period, tenant requests that have just arrived will not be served in this period but will be served in the next period, as in the literature [23], [24].…”

“…For the problem, the ordering of these requests is important and affects the solutions. In this paper, the most data volume first (MDVF) algorithm is applied [24]. MDVF is a descending ordering based on the data volume in the requests.…”

“…MDVF is a descending ordering based on the data volume in the requests. It can provide a better approximation solution compared to the ordering of first-come-first-served [24].…”

A Learning-Automata-Based Congestion-Aware Scheme for Energy-Efficient Elastic Optical Networks

“…The need for bandwidth is growing more than ever before as technology evolves [1] and high bandwidth-dependent applications have increased at unprecedented level. Future needs will also be driven by emerging capacity-demanding applications, including autonomous vehicles, the internet of things, high bandwidth enhanced video and virtual reality.…”