W Wang Miao

SDN-enabled OPS with QoS guarantee for reconfigurable virtual data center networks

Optical packet switching (OPS) can enhance the performance of data center networks (DCNs)by providing fast and large-capacity switching capability. Benefiting from the software-defined networking (SDN) control plane, which could update the look-up-table (LUT) of the OPS, virtual DCNs can be flexibly created and reconfigured. In this work, we have implemented and assessed an SDN-based control framework for an OPS node, where the OpenFlow protocol has been extended in support of the OPS switching paradigm. Application flows are switched by the OPS at submicrosecond hardware speed, decoupled from the slower (millisecond timescale) SDN control operation. By the DCN infrastructure provider, the virtual networks become directly programmable with the abstraction of the underlying OPS node. Experimental results validate the successful setup of virtual network slices for intra-data center interconnect and quality of service (QoS) guarantee for high-priority application flows. Data plane resources are efficiently shared by exploiting statistical multiplexing. In addition, the capability of exposing per-port OPS traffic statistics information to the SDN controller enables the implementation and experimental validation of load balancing algorithms to improve the QoS performance.

Novel flat data center network architecture based on optical switches with fast flow control

We propose a novel flat and scalable data center network (DCN) architecture based on fast (nanosecond) distributed buffer-less optical switches and efficient optical flow control. The proposed DCN architecture scales as the square of the port count of the optical switches. In order to investigate the performance of the proposed architecture, the system operation of an electronic Top-of-the-Rack (ToR) and the optical switch is fully described, and all functional subsystems are modeled. The performance in terms of DCN scalability, average latency, packet loss, normalized throughput of the network, and electronic buffer size of the ToR is numerically assessed for a medium-size data center supporting 5760 servers and a large-size data center connecting 100 000 servers. Considering a traffic pattern with high inter-cluster (40%) traffic distribution and buffer size of 40 KB, the results report an end-to-end latency of less than 8.1 μs (including retransmission) and a packet loss <; 10-7 under a load of 0.4 for the large-size data center. Moreover, we provide a preliminary experimental validation of the DCN by using a 4 × 4 optical switch prototype showing dynamic switching at 40 Gb/s and error-free operation with less than 1.5 dB penalty for the longest path.

Lightness: A Function-Virtualizable Software Defined Data Center Network With All-Optical Circuit/Packet Switching

Modern high-performance data centers are responsible for delivering a huge variety of cloud applications to the end-users, which are increasingly pushing the limits of the currently deployed computing and network infrastructure. All-optical dynamic data center network (DCN) architectures are strong candidates to overcome those adversities, especially when they are combined with an intelligent software defined control plane. In this paper, we report the first harmonious integration of an optical flexible hardware framework operated by an agile software and virtualization platform. The LIGHTNESS deeply programmable all-optical circuit and packet switched data plane is able to perform unicast/multicast switch-over on-demand, while the powerful software defined networking (SDN) control plane enables the virtualization of computing and network resources creating a virtual data center and virtual network functions (VNF) on top of the data plane. We experimentally demonstrate realistic intra DCN with deterministic latencies for both unicast and multicast, showcasing monitoring, and database migration scenarios each of which is enabled by an associated network function virtualization element. Results demonstrate a fully functional complete unification of an advanced optical data plane with an SDN control plane, promising more efficient management of the next-generation data center compute and network resources.

Towards petabit/s all-optical flat data center networks based on WDM optical cross-connect switches with flow control

Scaling the capacity while maintaining low latency and power consumption is a challenge for hierarchical data center networks (DCNs) based on electrical switches. In this work we present a novel all-optical flat DCN architecture OPSquare that potentially addresses the scaling issues by employing parallel intra-/inter-cluster switching networks, distributed fast WDM optical cross-connect (OXC) switches, and a novel top-of-rack (ToR) switch architecture. The fast (nanoseconds) WDM OXC switches allow flexible switching capability in both wavelength and time domains and statistical multiplexing. The OPSquare DCN performance targeting Petabit/s capacity has been thoroughly assessed. First the packet loss, latency, throughput, and scalability are numerically investigated under realistic data center traffic model. Results indicate that when scaling the DCN size up to 1024 ToR switches, a packet loss ratio below 10-6 and a server end-to-end latency lower than 2 μs can be guaranteed at load of 0.3 with limited 20 kB buffer. Then, the experimental evaluation of the DCN by employing 4 × 4 OXC prototypes shows multi-path dynamic switching with flow control operation. The case deploying 32 × 32 and 64 × 64 OXC switches connecting 1024 and 4096 ToRs are emulated and limited performance degradation has been observed. The potential of switching higher-order modulation and waveband signals further proves the suitability of OPSquare architecture for Petabit/s and low-latency DCN by using optical switches with moderate radix.

On the cost, latency, and bandwidth of LIGHTNESS data center network architecture

We investigate and compare the cost, bandwidth, and latency of the LIGHTNESS data center network (DCN) architecture adopting optical switching technologies with current DCN architectures based on electrical switching technologies. The studies include the cost and latency models of all the required opto-electronics (electronic switches, transceivers, cables) and optical (circuit and packet switches) components to build the DCN architectures. Numerical results show that the LIGHTNESS DCN architecture has the lowest cost due to the huge reduction of expensive transceivers. Moreover, the flat LIGHTNESS DCN architecture outperforms the multi-stage electronic DCN architecture in terms of latency, while the bisection bandwidth can be dynamically and flexibly reconfigured according to the required applications.

Performance assessment of optical packet switching system with burst-mode receivers for intra-data center networks

We investigate the performance of a burst-mode receiver in an optical packet switching system. Experimental results indicate that a preamble of 25.6ns allows error-free operation of 10Gb/s asynchronous switched packets with 8dB dynamic range and 25ns minimum guard-time.

Scalable and Low Latency Optical Packet Switching Architectures for High Performance Data Center Networks

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.

Performance evaluation of Optical Packet Switches on high performance applications

This paper analyzes the performance impact of Optical Packet Switches (OPS) on parallel HPC applications. Because these devices cannot store light, in case of a collision for accessing the same output port in the switch only one packet can proceed and the others are dropped. The analysis focuses on the negative impact of packet collisions in the OPS and subsequent re-transmissions of dropped packets. To carry out this analysis we have developed a system simulator that mimics the behavior of real HPC application traffic and optical network devices such as the OPS. By using real application traces we have analyzed how message re-transmissions could affect parallel executions. In addition, we have also developed a methodology that allows to process applications traces and determine packet concurrency. The concurrency evaluates the amount of simultaneous packets that applications could transmit in the network. Results have shown that there are applications that can benefit from the advantages of OPS technology. Taking into account the applications analyzed, these applications are the ones that show less than 1% of packet concurrency; whereas there are other applications where their performance could be impacted by up to 65%. This impact is mostly dependent on application traffic behavior that is successfully characterized by our proposed methodology.

Petabit/s data center network architecture with sub-microseconds latency based on fast optical switches

We propose a novel flat Petabit/s optical DCN based on fast optical switching systems. Demonstration with 4×4 switch prototype shows dynamic switching operation at 40Gb/s with <;10-5 packet loss and 640ns latency with 0.4 load.

Experimental assessment of an SDN-based control of OPS switching nodes for intra-data center interconnect

An SDN-based control framework for OPS switches is experimentally assessed. The OpenFlow messages enabling control communication between SDN controller and OPS switch are validated. Dynamic OPS virtual network slices for intra-data center interconnect are successfully setup and monitored.