Search

To cope with the capacity increase and diversification of information services, enhancing efficiency, flexibility, and agility in optical transport networks is becoming increasingly important. Quick reconfiguration of the optical physical layer, such as changing optical node architectures or optical physical network topologies, provides additional adaptability to requirement or environment changes. This paper proposes a new topology description scheme to support optical physical layer resource management at the granularity of optical functionality blocks in a machine-processible manner. The proposed scheme specifies intra-node structures as well as inter-node fiber connections (links). Furthermore, different switching functionalities of individual optical components are specified in a common format using integer linear programming (ILP) formulas. The ILP formulas are described in a machine-readable GNU MathProg modeling language so as to be directly introduced to the path computation mechanisms. Based on the proposed topology description scheme, a path computation engine named PathFinder is prototyped and demonstrated for optical networks consisting of various optical components with different switching functionalities. The computational feasibility of PathFinder is evaluated through numerical experiments in terms of elapsed time for path computation and the proposed scheme is successful while retaining reasonable time ranges. Operating topology and node architecture changes based on the proposed scheme are also successfully demonstrated over a multi-granular hierarchical optical network testbed with real hardware. [ABSTRACT FROM AUTHOR]

A novel method for estimating energy consumption of networks at the national scale is presented. Accurate estimations of network energy consumption are increasingly important because of continuously increasing network traffic and the need to assess emerging energy technologies to achieve overall energy efficiency. However, direct estimations are not practical because of the complex, widely diverse, and undisclosed configurations of actual networks. The impact of each network's energy overhead, which is the energy consumed above that required to accommodate network traffic, on the energy consumption associated with these complex configurations must be inferred. In the proposed method, the energy overhead is quantitatively evaluated and introduced as overhead factors by comparing the energy consumption estimated from network configuration models (bottom-up methods) with reports of actual energy consumption (top-down methods). This proposed “unified” method is capable of long-term predictions of future technology trends, including network architecture changes. In this paper, we consider different overhead factors that serve as fitting parameters for different network areas, and demonstrate the procedure for determining each of these parameters through an estimation of the fixed broadband Internet in Japan. This unified method consistently estimates the long-term evolution of energy consumption from 2000 to 2030. [ABSTRACT FROM PUBLISHER]

In this paper, ‘Martini,’ a network interface controller chip for our original network called RHiNET, is described. Martini is designed to provide high-bandwidth and low-latency communication with a small overhead. To obtain high-performance communication, protected user-level zero-copy remote direct memory access (RDMA) communication functions are completely implemented by a hardwired logic. Also, to reduce the communication latency efficiently, we have proposed programmed I/O (PIO)-based communication mechanisms called ‘On the fly (OTF)‘ and have implemented them on Martini. The evaluation results show that Martini connected to a 64-bit/66-MHz peripheral component interconnect (PCI) bus achieves a 470-Mbyte maximum bidirectional bandwidth and a 1.74-μsec minimum latency on host-to-host memory copying. [ABSTRACT FROM AUTHOR]

Ethernet has been used for connecting hosts in PC clusters, besides its use in local area networks. Although a layer-2 Ethernet topology is limited to a tree structure because of the need to avoid broadcast storms and deadlocks of frames, various deadlock-free routing algorithms on topologies that include loops suitable for parallel processing can be employed by the application of IEEE 802.1Q VLAN technology. However, the MPI communication libraries used in current PC clusters do not always support tagged VLAN technology; therefore, at present, the design of VLAN-based Ethernet cannot be applied to such PC clusters. In this study, we propose a switch-tagged routing methodology in order to implement various deadlock-free routing algorithms on such PC clusters by using at most the same number of VLANs as the degree of a switch. Since the MPI communication libraries do not need to perform VLAN operations, the proposed methodology has advantages in both simple host configuration and high portability. In addition, when it is used with on/off and multispeed link regulation, the power consumption of Ethernet switches can be reduced. Evaluation results using NAS parallel benchmarks showed that the performance of the topologies that include loops using the proposed methodology was comparable to that of an ideal one-switch (full crossbar) network, and the torus topology in particular had up to a 27 percent performance improvement compared with a tree topology with link aggregation. [ABSTRACT FROM AUTHOR]