A. Asensio

Towards a carrier SDN: an example for elastic inter-datacenter connectivity

We propose a network-driven transfer mode for cloud operations in a step towards a carrier SDN. Inter-datacenter connectivity is requested in terms of volume of data and completion time. The SDN controller translates and forwards requests to an ABNO controller in charge of a flexgrid network.

Study of the centralization level of optical network-supported Cloud RAN

Radio Access Network (RAN) costs savings are expected in future Cloud RAN (C-RAN). Differently from traditional distributed RAN architecture, in C-RAN Remote Radio Heads (RRH) from different sites can share baseband processing resources from virtualized Base Band Unit (BBU) pools placed in few central locations (CO). Due to the stringent requirements of the several interfaces needed in C-RAN, optical networks have been proposed to support it. In this paper, we present the C-RAN Capital Expenditures (CAPEX) minimization problem to decide which COs should be equipped and the equipment to be installed. Several problem instances considering different Long Term Evolution-Advanced (LTE-A) configurations are solved to study the impact of the centralization level C-RAN in terms of both, CAPEX and Operational Expenditures (OPEX). Compared to the maximum centralization level, results showed remarkable costs savings when a lower level of centralization is considered.

Scalability of telecom cloud architectures for live-TV distribution

A hierarchical distributed telecom cloud architecture for live-TV distribution exploiting flexgrid networking and SBVTs is proposed. Its scalability is compared to that of a centralized architecture. Cost savings as high as 32 % are shown.

An elastic networks OMNeT++-based simulator

An increasing number of studies about flexgrid-based elastic optical networks are currently being carried out by industry as well as academic research teams. As access to real test-beds is limited, there is a lack of effective tools that emulate realistic scenarios for testing algorithms and network architectures. In this paper, we present a simulation framework based on an active stateful Path Computation Element (PCE) with Global Concurrent Optimization (GCO) for flexgrid optical networks. The simulator is fully configurable and allows implementing and testing new algorithms and architectures easily and quickly.

Managing transfer-based datacenter connections

Creating datacenter federations allows balancing workloads accommodating spikes in cloud demand while reducing the extensive energy consumption of datacenters. This allows datacenter operators to reduce cost and increase their revenue from using under-utilized resources. To implement those elastic operations, datacenters need to be connected through a network providing huge capacity and guaranteeing adequate resource availability when it is required. Those features can be covered by the flexgrid optical technology provided with an intelligent control plane. For the latter, in this paper we rely on the application-based network operations (ABNO) architecture currently under standardization in the IETF. To increase the availability of the desired capacity, a carrier software defined network (SDN) is placed on top of ABNO. On this architecture, cloud resource managers request data transferences using an application-oriented semantic, which includes data volume and required completion time. Those requests are transformed into connection requests by the carrier SDN. To increase resource availability for incoming requests, the carrier SDN can perform elastic operations on already established connections supporting transferences provided that the committed completion time is ensured. Requests are routed and resource allocation is scheduled, which becomes the routing and scheduled spectrum allocation (RSSA) problem. An integer linear programming model is proposed, and an algorithm for solving the RSSA problem in realistic scenarios is designed. Results showing remarkable gains in the amount of data transported motivate its use for inter-datacenter traffic.

Dynamic virtual network connectivity services to support C-RAN backhauling

Traditional distributed radio access networks (RANs) are evolving toward centralized architectures, with the aim of satisfying in a cost-effective manner the forecast traffic growth that future mobile networks will need to support. Specifically, the cloud RAN (C-RAN) architecture has shown that it can alleviate to some extent the everincreasing total cost of ownership in mobile networks. The current trend in C-RAN is to separate remote radio heads with radio frequency (RF) functions and baseband unit (BBU) gathering baseband processing. This functional split allows the RF modules to be kept close to the antennas while placing BBUs at centralized locations so they can be shared among different sites and even be virtualized. However, some issues still need to be addressed in future mobile networks, especially due to the dynamicity of services and the strict constraints imposed by the interfaces needed. In fact, connectivity reconfiguration for X2 and S1 backhaul interfaces needs to be provided as an all-or-nothing request to enable mobile resource reconfiguration in a geographical area. In view of this, in this paper we propose dynamic customer virtual network (CVN) reconfiguration to be supported in metro and core network segments. Such CVN requests must include quality of service constraints to ensure specific delay constraints, as well as bitrate guarantees to avoid service interruption. A mathematical formulation and heuristic algorithm are presented for the CVN reconfiguration problem, and exhaustive simulation results study its performance in realistic scenarios.

Dynamic customer virtual network reconfiguration with QoS constraints and bandwidth guarantees

An algorithm for Customer Virtual Network (CVN) reconfiguration with QoS constraints and bandwidth guarantees is presented. An MPLS virtual network supporting CVNs is pre-planned or dynamically reconfigured. Exhaustive simulations results show 19% CAPEX savings when the MPLS network is reconfigured.

Carrier SDN for flexgrid-based inter-datacenter connectivity

Managing inter-datacenter (DC) connectivity to transport increasing amounts of data and optimize DC performance and costs becomes a challenge not only for DC operators but also for network operators. Current inter-DC connections are configured as static big fat pipes, resulting in large bitrate over-provisioning thus increasing operational costs for DC operators. Furthermore, because of the nature of inter-DC traffic, which highly varies during the time, network operators cannot share such connections with other customers, and network resources may be underutilized. Therefore, an increasing number of studies are being carried out with the aim to address different challenges arising from inter-DC connectivity requirements. This paper summarizes connectivity models and results showing remarkable advantages of considering carrier software defined network (SDN) and elastic operations for managing inter-DC connections.

Using spectrum management techniques to differentiate services in Elastic Optical Networks

Elastic Optical Networks (EONs) have emerged as a promising solution to cope with the growth and heterogeneity of networks data traffic. Nevertheless, several problems arise in dynamic scenarios, being spectrum fragmentation the most important one. Under a fragmented spectrum, connections with large bandwidth experience higher problems in finding enough free bandwidth to be set-up. In this work, creating spectrum sub-bands dedicated to different types of connections is used to provide different blocking probability values to these sets of connections. By properly adjusting the design and use of these sub-bands, it is possible to tune the network performance attaining different blocking probability (BP) values as a function of the spectrum band division and the priority given to the different connection requests sizes. The purpose of this paper is to demonstrate that different BP values can be obtained for different classes of traffic which, in turn, are assigned to different connections sizes. The strategies proposed are assessed by means of simulation.

Minimizing energy costs in federated datacenters under uncertain green energy availability

The cost of energy represents, by far, the largest fraction of total operational expenditures that datacenter operators ought to face. For this very reason, several studies have focused on evaluating how such energy costs can be reduced and on quantifying that reduction; using green energy sources (e.g. solar) that can be generated by installing infrastructures nearby datacenters is clearly an interesting option. Assuming that green energy is available, workloads consolidation in those datacenters with the highest amount of self-generated energy allows reducing remarkably the consumption of brown energy. Workload management is of paramount importance to increase green energy consumption in the context of distributed datacenters. In that scenario, a centralized and orchestrated operation leads to large energy cost savings. To this end, we firstly present a model to estimate the amount of green energy produced in each location as a function of the specific time period and the expected weather conditions. Next, the problem of minimizing energy costs by properly placing workloads in federated datacenters under uncertainty in the availability of green energy in each location is faced using stochastic programming techniques. Illustrative numerical results validate the usefulness of the proposed stochastic approach.