Current Memory-Space-Memory Clos switching fabric exhibits poor performance under unbalanced traffic. To eliminate the internal congestion within the switching fabric, by taking buffers into the internal links, this p...
详细信息
ISBN:
(纸本)9781424445530
Current Memory-Space-Memory Clos switching fabric exhibits poor performance under unbalanced traffic. To eliminate the internal congestion within the switching fabric, by taking buffers into the internal links, this paper proposes an alternative Fully Buffered three-stage Clos switching fabric - FBC, which not only provides perfect throughput performance, but also exhibits betterdelay performance in heavy load by storing cells at the internal buffers. To employ an efficient and simple scheduling process, we provide a scheduling algorithm for FBC which is nameddistributedround-robin Scheduling algorithm (drrS). This algorithm takes advantages of high-efficiency of round-robin and is distributed in implementation, which makes the drrS prone to implement in hardware. At last, the simulation results prove that drrS is more effective.
As more and more mission critical service are now transported over high-speed networks, it becomes an important issue to make use of the relativity of multiple link/node failures and improve transport performance for ...
详细信息
As more and more mission critical service are now transported over high-speed networks, it becomes an important issue to make use of the relativity of multiple link/node failures and improve transport performance for high-speed networks. In this paper, we propose a novel algorithm called low cost an S disjoint paths algorithm (LCSd for short) to establish a SrLG disjoint active path and backup path pair for network protection. Based on the TF algorithm, LCSd algorithm introduces the number of links which relate to the SrLG as a new constraint. Through link weight adjustment before the second time SPF calculate, the backup path can avoid the risk sharing with active path. Theoretical analysis and experiment results have shown that LCSd algorithm resolved the trap problem and path pair total cost non-optimal problem under SrLG disjoint constraint. Compared with rF and TF algorithms, LCSd algorithm is significantly superiority in the successful ratio to find feasible solution and can guarantee minimum total cost of active and backup path.
Although several available bandwidth measurement tools exist, they usually require access at both ends of the measured path. In this paper, we present ABuGET, a single-ended tool for available bandwidth estimation aga...
详细信息
Although several available bandwidth measurement tools exist, they usually require access at both ends of the measured path. In this paper, we present ABuGET, a single-ended tool for available bandwidth estimation against non-cooperative hosts. Comparing to other single-ended tools based ONE-dELAY mostly. Our measurement tool focus on utilizing ITPr (the improved PTr) algorithm to realize and exploiting TCP protocol options to initiate probing traffic. We describe the design of ABuGET. The comparison of ABuGET to Pathload (the most popular tool measuring the available bandwidth) used in simulation environments showed that our tool can provide better quality of measurements in non-cooperative environments.
In this paper, we provide a non route-calculating MPLS LdP fast protection switching scheme called LdP NrC-FPS scheme which is efficient in transient multi-failures in MPLS networks. This scheme makes full use of vari...
详细信息
In this paper, we provide a non route-calculating MPLS LdP fast protection switching scheme called LdP NrC-FPS scheme which is efficient in transient multi-failures in MPLS networks. This scheme makes full use of various working modes and configurable loop detection of LdP protocol.. When network failures happen, the scheme need not to compute routing again. The scheme uses Path Vector TLV and Hop Count TLV in LdP mapping message to avoid loop in transient multi-failure scenario. Furthermore, OAM mechanism has been introduced into LdP NrC-FPS scheme for accelerating fault recovery process. Compared with traditional fast rerouting strategies in IP and MPLS layer, it reduces the failure recovery time of rerouting and has the advantage of more efficient network resource utilization. Performance analysis shows that LdP NrC-FPS scheme outperforms than traditional IP fast reroute scheme and MPLS protection switching scheme at loop ratio , path unreachable ratio and control message overhead aspects.
Forwarding and control element separation (ForCES) architecture network devices can satisfy the requirements of reconstruction for the new generation of the Internet better. This paper proposes a novel synthetic netwo...
详细信息
Forwarding and control element separation (ForCES) architecture network devices can satisfy the requirements of reconstruction for the new generation of the Internet better. This paper proposes a novel synthetic network management model based on ForCES. This model regards the device under management (dUM) as forwarding element (FE) in ForCES architecture and uses FE model to abstract the definition of management function block (MFB), dUM communicates with network management controller using ForCES. This model provides an efficient and flexible method for synthetic network management of diverse dUMs.
Understanding the topology of large scale IPv6 networks is essential to the procurement of good architectural design decisions, particularly with respect to address allocation anddistribution schemes. The formerdist...
详细信息
ISBN:
(纸本)9781424442348
Understanding the topology of large scale IPv6 networks is essential to the procurement of good architectural design decisions, particularly with respect to address allocation anddistribution schemes. The formerdistributed topology discovery applications deploy multiple probing hosts throughout the network. This approach relies on the ability to deploy probing hosts throughout the network, which in many cases is not possible. Source routing support in IPv6 routers is a mandated feature and hence mostly enabled. With respect to simple traceroute, source routing allows one to discover paths between arbitrary pairs of network nodes, but it is such difficult to avoid probing redundancy with a null probing stopping set in the initial phrase of discovery. This paper presents two probing systems: VdPS (virtual distributed Probing system) and SrPS (Source routing Probing system). Firstly, VdPS discovers the BTI (Basic Topology Information) of IPv6 networks. Secondly, SrPS selects seeds from the BTI. In the processing of SrPS, it is very effective in avoiding probing redundancy rely on the probing stopping set initiated by the BTI. Finishing the implementation of SrPS, the coverage of probing is increased greatly. We also present some initial results from probing the CErNET2, currently the largest public IPv6 network of China. The results illustrate the effectiveness of coverage and avoiding probing redundancy.
This paper presents the M/G/1 multi-class open queuing networks performance model of IXP2400-based VoIP SPS (signaling parse system). We use UML activity diagrams and function module diagrams to describe this VoIP SYS...
详细信息
ISBN:
(纸本)9781424442317
This paper presents the M/G/1 multi-class open queuing networks performance model of IXP2400-based VoIP SPS (signaling parse system). We use UML activity diagrams and function module diagrams to describe this VoIP SYS, andderive the performance model. The resulting model could make statistic analysis of IXP2400-based VoIP SPS with First-Come-First-Seine (FCFS) scheduling policy and mix traffic. We use the approximate calculation of iterative decomposition algorithm to solve this model, and present the formulas of the utilization, delay, etc. The theoretical estimates are in excellent consistence with our testing data, this shows the effectiveness of our performance evaluating method.
Stream merging techniques utilize multicast to minimize the serverresource consumption. However, how waiting requests are scheduled for service in the context of stream merging has not been studied extensively. The I...
详细信息
ISBN:
(纸本)9781424424238
Stream merging techniques utilize multicast to minimize the serverresource consumption. However, how waiting requests are scheduled for service in the context of stream merging has not been studied extensively. The Improved Maximum Urgency First (IMUF) scheduling algorithm for video servers applying stream merging is proposed. IMUF considers multiple attributes of the video waiting queue. Three heuristic implementation schemes are presented: IMUF-I, IMUF-P and IMUF-M. The proposed implementations are compared and evaluated through extensive simulation. results show that IMUF can yield significant performance benefits in terms of defection rates, average access delay and fairness.
Many multicast video streaming algorithms have been proposed to provide scalable streaming media services, but they are designed for homogeneous clients, i.e., users are assumed to have the same receiving bandwidth. T...
详细信息
ISBN:
(纸本)9781424424238
Many multicast video streaming algorithms have been proposed to provide scalable streaming media services, but they are designed for homogeneous clients, i.e., users are assumed to have the same receiving bandwidth. The performance of these state-of-the-art algorithms will degrade significantly in user heterogeneous environment. A scheme, called the Enhanced Patching, is proposed to support user heterogeneity. Both the optimal threshold window and the average bandwidth are analyzed. A simple but efficient analytical model is also proposed to evaluate the distribution of the capacity demanded by the clients. The numerical results show the feasibility of the proposed scheme.
In this paper we describe our experience in performance evaluation of CMP-based VoIP SPS (signaling parse system). We also present a performance model of the system using GI/G/m-FCFS multi-class open QNs(queuing netwo...
详细信息
ISBN:
(纸本)9780769531311
In this paper we describe our experience in performance evaluation of CMP-based VoIP SPS (signaling parse system). We also present a performance model of the system using GI/G/m-FCFS multi-class open QNs(queuing networks). Anddeal with approximate performance analysis of open non-product-form QNs, based on the method of decomposition. Our approach predicts the major performance evaluation parameters of the system under the workloads and also identifies bottlenecks in the system. Performance results are inserted back into the QN model as initial values associated to the relevant model elements. The theoretical estimates are shown to be in excellent consistence with testing results. It shows the effectiveness of the proposed model. This work explores performance issues of CMP-based applications by means of performance modeling. In this way we may get an immediate feedback on the system performances. The use of QN model allows us to achieve better multi-core application performance.
暂无评论