A distributed and scalable optical packet switch architecture for data centers

We review an optical flat datacenter network (DCN) based on scalable optical switching system with port-count independent low latency. Generation/reconfiguration of virtual DCN by utilizing statistical multiplexing offered by the optical switching system is demonstrated.

As the scale of data centers and cloud computing applications increases, data center networks play a critical role in meeting the huge communication bandwidth requirement of such applications. The scalability of conventional electronic data center networks is limited by wiring complexity and reaching distance of links under fixed power budget. To overcome this problem, in this paper we propose a nonblocking hybrid switch architecture, called NEO (Nonblocking Electronic and Optical), which is able to provide nonblocking interconnections for as many as 1,000,000 servers in a data center. NEO maintains electronic interconnections for intra-pod networks, while providing interpod interconnections by optical core switches, which not only increases the scalability of the switch architecture, but also lowers the switch cost and power consumption compared to other existing optical switch architectures. We also design a packet scheduler for NEO, which adopts a credit flow control mechanism and a parallel scheduling algorithm to avoid packet loss, and provide low communication latency. Our simulation results demonstrate that NEO achieves very low average packet delay compared to other existing optical switching architectures under various traffic patterns.

In this paper, we propose an analytical model to evaluate the power consumption in the switching fabric of a bufferless shared-per-wavelength (SPW) optical packet switch architecture in which one bank of wavelength converters (WC) is dedicated to each wavelength. We assume that both optical gates and WCs are realized in semiconductor optical amplifier technology. In our evaluation, we account for the power consumption of the current drivers needed to both controlling the used active devices and supplying the thermoelectric coolers. SPW allows for a complexity reduction of the spatial switching matrix that leads to reduced power consumption with respect to other switching architectures. Results show the effectiveness in terms of consumed power of the considered architecture with respect to the shared-per-node reference architecture, where a fully sharing strategy of WCs is adopted. The main results show that SPW allows us to reduce the power consumption in the order of 26% for offered traffic equal to 0.6. The obtained results also show how the fabric switching of the SPW optical packet switch consumes much less power per gigabits per second carried than the one of a typical commercial core router.

Current electronic broadband switch matrices employ discrete semiconductor devices or VLSI chips. Isolation between ON- and OFF-states degrades and crosstalk between input and output lines of these electronic switch matrices increases at high operation frequencies. These will limit electronic switch matrices to low-frequency or low-bit-rate operations. In addition, optical-to-electrical (electrical-to-optical) conversions at the input (output) ends are required for these switch matrices to be used in a fiber environment. One way to achieve broadband, low-crosstalk, and high-isolation switching and to eliminate E/O and O/E conversions is to use optical switches. This paper describes an optical broadband switching architecture using optical/optoelectronic logic gates. Input and output signals are optical. No E/O or O/E conversions are needed for this architecture. The optical/optoelectronic logic gates that can be employed in this architecture are reviewed. Performance of this architecture is evaluated by calculating the crosstalk, isolation, insertion loss, switch matrix size, drive power, and switching speed. Technologies needed for hybrid and monolithic implementations of this architecture are also identified.

We demonstrate end-to-end 25Gb/s true optical packet switching featuring burst-mode reception with $\text{Hipo}\lambda\text{aos}$ Optical Packet Switch architecture. Error-free performance at 10−9 was obtained for all validated port-combinations.

In order to find out the tradeoff between recovery time and resource overhead when a link failure occurs, we propose a software-defined based fast restoration scheme for elastic optical datacenter networks, called precomputation based restoration path (P-RP). By extending the controller functionality and OpenFlow protocol in software defined networking (SDN) technology, we establish a novel elastic optical inter-datacenter network architecture which can quickly and accurately converge the network state information with a global view for failure recovery. Based on this architecture, the P-RP performs path computation and bandwidth resource selection for restoration before a failure occurs. The experimental results demonstrate that the proposed scheme can achieve fast recovery with low blocking probability while maintaining high spectrum efficiency. Compared with existing restoration schemes, average recovery time is improved by up to 28%.

We propose an optical interconnection architecture in both economy and energy efficiency called as MMTD net (MEMs based Multi-Tier Data center network). Numerical results show that this architecture is highly scalable, with low blocking probability and latency.

We present a field-programmable gate array (FPGA)-based label processor for in-band optical labels with a processing time independent of the number of label bits. This allows for implementing an optical packet switching architecture that scales to a large port count without compromising the latency. As a proof of concept, we have employed an FPGA board with 100 MHz clock to validate the operation of the label processor in a 160 Gb/s optical packet switching system. Experimental results show successful three label bits processing and 160 Gb/s packets switching with 1 dB power penalty and 470 ns of latency. Projections on the label processor performance by using more powerful FPGAs indicate that 60 label bits ( optical addresses) can be processed within 31 ns.

To cope with ever-increasing traffic demands in transport networks, all-optical switching is currently perceived as a solution to remove bottlenecks imposed by O/E/O conversions. Optical label switching accomplishes packet switching using the optical-labels. It simplifies the design of switching devices in the optical domain and speeds up packet forwarding significantly. In this paper, we propose a novel design of optical switches, and present a performance analysis to show its in-depth behavior. The analysis results demonstrate the effectiveness of the proposed switching architecture.

Optical packet switching is the next generation switching technique in conjunction with mature electronic switching. In optical packet switching, the core router plays an important role in the routing and buffering of packets. In recent past, AWG based optical router are proposed and heavily investigated due to its very low insertion loss and very simple wavelength dependent routing pattern. In this paper, a further modification in an AWG based optical packet switch is proposed. These modifications improve the signal reception quality at the output. Power budget analysis is presented to evaluate the buffer capacity. As a physical layer parameters, the minimum switch power is evaluated using the mathematical model and at the BER level of 10−9 the maximum buffering capacity is evaluated. Finally, simulation results are presented in terms of network layer parameters, packet loss probability and average delay. The packet loss probability is as low as 10−7 well under acceptable limits.

This paper compares selected Optical Packet Switching architectures that use the wavelength conversion technique to solve the packet contention problem. The architectures in question share wavelength converters, which are needed to wavelength translate arriving packets. This paper focuses on two architectures: the Shared Per Output Line (SPOL) and the Shared Per Input Line (SPIL) architectures, in which the wavelength converters are shared per output and input fiber respectively. The performance of the proposed architectures is evaluated for all the balance/unbalance combinations of input/output traffic. Packet loss probability is expressed as a function of the number of wavelength converters used, by means of analytical models validated by simulations. The results obtained show that the SPIL architecture, when compared to the SPOL architecture, allows for greater economies in terms of number of wavelength converters needed. While the performance of the two architectures tends to have similar values in a scenario with unbalanced input traffic and balanced output traffic, in unbalanced output traffic scenarios the SPIL architecture requires about 50% less wavelength converters than the, SPOL architecture does, for a given packet loss probability.

In this paper, we propose a new ant-based algorithm for the dynamic routing and wavelength assignment (RWA) problem in optical WDM networks under the wavelength continuity constraint. Unlike conventional approaches, which usually require centralized global network information, our new RWA algorithm constructs the routing solution in a distributed manner by means of cooperative ants. To facilitate the ants’ foraging task, we adopt in our algorithm a probabilistic routing table structure for route selection. The new algorithm is highly adaptive in that it always keeps a suitable number of ants in the network to cooperatively explore the network states and continuously update the routing tables, so that the route for a connection request can be determined promptly by the current states of routing tables with only a small setup delay. Some new schemes for path scoring and path searching are also proposed to enhance the performance of our ant-based algorithm. Extensive simulation results upon three typical network topologies indicate that the proposed algorithm has a very good adaptability to traffic variations and it outperforms both the fixed routing algorithm and the promising fixed–alternate routing algorithm in terms of blocking probability. The ability to guarantee both a low blocking probability and a small setup delay makes the new ant-based routing algorithm very attractive for both the optical circuit switching networks and future optical burst switching networks

As optical switching and multiplexing architectures evolve, optical multicast keeps on being a capability advertised by a few core and transport optical equipment vendors. However, to the date, not much use of this service has been made by operators. This work addresses the reasons behind this fact, identifies the key technical design factors in a multicast-capable network and the challenges to face for an effective use of multicast at the optical layer.

A novel optical implementation method of 4 X 4 polarization-independent bi-directional fiber optical switch in free-space optical architecture is presented in this paper. This 4 X 4 fiber optical switch is based on elementary 2 X 2 optical switch module. The 4 X 4 optical switch constituted by polarization beam splitters (PBS), 1/4 waveplates (QWP), polarization light modulator (PLM), right angle prism (RAP), and total reflection mirror (TR). Operation is independent by input signals of polarization of the optical beams. This new kind of configuration of the optical switch grants the features of less optical components, high compactness, low optical interchannel crosstalk, fast switching speed, polari insensitivity, and easiness to optical assembly. A matrix description is deduced with respect to this 4 X 4 fiber optical switch architecture.

Multi-wavelength switching in IP optical nodes adopting different buffering strategies