Passive optical networks (PON) technology is increasingly becoming an attractive solution in modern data centres as it provides energy efficient, high capacity, low cost, scalable and flexible connectivity. In this paper we report the implementation of a PON based data centre architecture that provides high resilience and high speed interconnections by providing alternative communication routes between servers in different racks. Each rack is divided into several groups of servers and connects to other racks and the Optical Line Terminal (OLT) through a set of server that acts as relay servers. We implement the switching and routing functionalities within servers using 4x10GE Xilinx NetFPGA, and demonstrate end-to-end communication using IP cameras live video streaming over up to 100 km optical connections through WDM nodes and the PON network.

Passive Optical Data Centre Networks have been developed due to the performance limitations in current data centres to provide high performance within data centre networks. An AWGR based passive optical network data centre architecture is evaluated using a TDM/WDM multiple access technique to provision the flow of traffic among the network efficiently. A Mixed Integer Linear Programming model is developed to optimise resource allocation in the architecture. Using WDM-TDM as a multiple access technique helps in solving issues such as oversubscription and congestion by allowing servers to make simultaneous transmissions of packets in different time slots. The results show that the provisioning / allocation of resource within the architecture is improved with improvements of up to 75% in resource utilisation. WDM-TDM PON-BASED DATA CENTRE ARCHITECTUREIn this architecture each PON Cell contains two AWGRs as shown in Figure.1, to provision full interconnection between the 4 racks. Each rack contains 32 servers, using 4 different wavelengths. A 1: N AWGR connects the racks and the OLT. The network interface card of each server contains an array of photo detectors for wavelength detection and tuneable lasers for wavelength selection [11].Inter-rack communication is done through the direct AWGR link or through the OLT. The AWGR selects the wavelength for transmission according to the location of the destination server, and servers communicate according to an AWGR wavelength routing interconnection map. The availability of alternative routes provides multi-path routing and enhances load balancing at high traffic load [11].

Passive optical networks (PON) technology has recently been proposed as a solution for scalability, energy efficiency, high capacity, low cost, flexibility and oversubscription issues in data centres. This paper experimentally demonstrates and discusses the implementation of a server centric PON based data centre architecture with high speed and reliability. The architecture is set up using a set of servers grouped into racks directly connected together and to the Optical Line Terminal (OLT) through gateway servers. The switching and routing functionalities have been embedded into servers using 4x10GE Xilinx NetFPGA. Flow continuity has been observed through live video streaming using IP cameras transmitting over up to 110 km optical connections through WDM nodes and the PON network.

Passive optical networks (PON) can play a vital role in data centres and access fog solutions by providing scalable, cost and energy efficient architectures. This paper proposes a Mixed Integer Linear Programming (MILP) model to optimize the placement of virtual machines (VMs) over an energy efficient WDM-TDM AWGR PON based data centre architecture. In this optimization, the use of VMs and their requirements affect the optimum number of servers utilized in the data centre when minimizing the power consumption and enabling more efficient utilization of servers is considered. Two power consumption minimization objectives were examined for up to 20 VMs with different computing and networking requirements. The results indicate that considering the minimization of the processing and networking power consumption in the allocation of VMs in the WDM-TDM AWGR PON can reduce the networking power consumption by up to 70% compared to the minimization of the processing power consumption.

The number of applications running in the cloud has dramatically increased in the past decade as well as the number of users accessing them. Data centres resources, architectures and conditions define the performance of the applications running on them. One of the main measures of the network efficiency is the latency, which can have a huge effect on resources utilisation, power consumption and the overall performance. In this paper, the performance of a fog network is evaluated by measuring the latency while running a facial recognition software. The network consists of two processing cells, a core network and a PON cell. The results show how network latency is affected by running the facial recognition software in the end-to-end network setup introduced in this paper.