Bandwidth guaranteed multicast scheduling for virtual output queued packet switches

Multicast enables efficient data transmission from one source to multiple destinations, and has been playing an important role in Internet multimedia applications. Although several multicast scheduling schemes for packet switches have been proposed in the literature, they usually aim to achieve only short multicast latency and high throughput without considering bandwidth guarantees. However, fair bandwidth allocation is critical for the quality of service (QoS) of the network, and is necessary to support multicast applications requiring guaranteed performance services, such as online audio and video streaming. This paper addresses the issue of bandwidth guaranteed multicast scheduling on virtual output queued (VOQ) switches. We propose the Credit based Multicast Fair scheduling (CMF) algorithm, which aims at achieving not only short multicast latency but also fair bandwidth allocation. CMF uses a credit based strategy to guarantee the reserved bandwidth of an input port on each output port of the switch. It keeps track of the difference between the reserved bandwidth and actually received bandwidth, and minimizes the difference to ensure fairness. Moreover, in order to fully utilize the multicast capability provided by the switch, CMF lets a multicast packet simultaneously send transmission requests to multiple output ports. In this way, a multicast packet has more chances to be delivered to multiple destination output ports in the same time slot and thus to achieve short multicast latency. Extensive simulations are conducted to evaluate the performance of CMF, and the results demonstrate that CMF achieves the two design goals: fair bandwidth allocation and short multicast latency.     

QoS-supported On-chip Communication for Multi-processors

We present a Quality of Service (QoS)-supported on-chip communication that increases the shared communication resources for multi-processor systems on chip. Time-critical embedded systems require tight guaranteed services in terms of throughput, latency etc. in order to comply to hard real-time constraints. Typically, guaranteed-service schemes require dedicated/reserved resources (i.e. links) for communication and thus suffer from low resource utilization. So improving the bandwidth utilization by using the unused bandwidth among the other competing transactions in a fair fashion is an important issue. To the best of our knowledge, we are presenting the first approach for on-chip communication that provides a high resource utilization under a transaction-specific, flexible communication scheme. It provides tight time-related guarantees through our bounded arbitration scheme considering the lower and the upper bounds for each type of transactions. We demonstrate its advantages by means of a complete MPEG4 video decoder case study analysis and achieve under certain constraints a bandwidth utilization of up to 100% and 97% on average with a guaranteed 100% bandwidth. Thus, we provide an on-chip communication scheme that provides high bandwidth utilization while providing tight guarantee. Large parts of this work have previously been published at the IEEE International Conference on Hardware/Software Codesign and System Synthesis (Codes+ISSS), 2006.     

定向天线卫星网络空间重用的带宽预留算法

基于无线网络中设计提供服务质量(quality of service,QoS)的路由协议是一项具有挑战性的工作,提出无线卫星网络中基于定向天线的服务质量保证的空间重用的带宽预留算法。卫星网络中许多音/视频会议、远程教育等重要应用,需要服务质量保证。无线网络中的定向天线技术提供了显著增加空间重用性的能力,提高无线网络中的数据传输效率。研究使用定向天线的无线卫星网络中基于时分多址的带宽预留算法,该算法给不同方向的地面终端分配相同时隙来提高带宽资源的空间重用性,提高通信效率。通过模拟实验分析研究证明,该算法在服务质量调用成功率、吞吐量和延迟方面有比较好的性能。     

Advance reservations for service-aware GMPLS-based optical networks

In service-oriented networks, cooperation between users, applications and network tends to come closer. With this objective, new architectures arise to cope with the upcoming demands. Of great importance is the definition of the proper interfaces and network services. One of these new services is known as the Advance Reservation Service. This service allows prior reservation of network resources in order to provide guaranteed client connections. The introduction of advance reservations in heterogeneous transport networks imposes the need to find new mechanisms that consider temporal characteristics of scheduled connections and their impact on immediate connections. In this paper, we investigate these issues, propose novel algorithms to improve overall network performance, and describe the implementation carried out over the ASON/GMPLS CARISMA testbed.     

A core stateless bandwidth broker architecture for scalable support of guaranteed services

We present a novel bandwidth broker architecture for scalable support of guaranteed services that decouples the QoS control plane from the packet forwarding plane. More specifically, under this architecture, core routers do not maintain any QoS reservation states, whether per-flow or aggregate. Instead, the QoS reservation states are stored at and managed by a bandwidth broker. There are several advantages of such a bandwidth broker architecture. Among others, it avoids the problem of inconsistent QoS states faced by the conventional hop-by-hop, distributed admission control approach. Furthermore, it allows us to design efficient admission control algorithms without incurring any overhead at core routers. The proposed bandwidth broker architecture is designed based on a core stateless virtual time reference system developed recently. This virtual time reference system provides a unifying framework to characterize, in terms of their abilities to support delay guarantees, both the per-hop behaviors of core routers and the end-to-end properties of their concatenation. We focus on the design of efficient admission control algorithms under the proposed bandwidth broker architecture. We consider both per-flow end-to-end guaranteed delay services and class-based guaranteed delay services with flow aggregation. Using our bandwidth broker architecture, we demonstrate how admission control can be done on a per domain basis instead of on a "hop-by-hop" basis. Such an approach may significantly reduce the complexity of the admission control algorithms. In designing class-based admission control algorithms, we investigate the problem of dynamic flow aggregation in providing guaranteed delay services and devise a new apparatus to effectively circumvent this problem. We conduct detailed analyses to provide theoretical underpinning for our schemes as well as to establish their correctness. Simulations are also performed to demonstrate the efficacy of our schemes.     

Statistical real-time channels on multiaccess bus networks

Real-time communication with performance guarantees is expected to become an important and necessary feature of future computer networks. In this paper, we present a scheme which can provide real-time communication services with both absolute and statistical performance guarantees on multiaccess bus networks for given input traffic characteristics and performance requirements. The proposed scheme reserves network bandwidth for real-time connections according to their needs. It also allows for independent addition and deletion of real-time connections while preserving existing guarantees. Our extensive simulation results for motion video communication have shown the proposed scheme to outperform the other well-known schemes     

无线多媒体网络中一种基于测量网络状态的动态呼叫接纳控制算法

提出了一种无线多媒体网络中基于测量网络状态的动态呼叫接纳控制算法.它区分了实时和非实时业务,在网络带宽资源不足时可通过降低非实时业务带宽确保实时业务呼叫连接的可靠性;还可根据当前网络状况调整预留带宽大小,使小区实时业务切换呼叫掉线率低于设定的门限值.大量仿真结果显示该算法具有低实时业务切换呼叫掉线率和与固定预留方案相当的带宽利用率,而只以略高的新呼叫阻塞率为代价,适合各种不同概率发生时实际应用的情况.     

An Approach to Quality of Service Management in Distributed Multimedia Application: Design and an Implementation

Most work related to quality of service (QoS) is concerned with individual system components, such as the operating system or the network. However, to support distributed multimedia applications, the entire distributed system must participate in providing the guaranteed performance levels. In recognition of this, a number of QoS architectures have been proposed to provide QoS guarantees. The mechanisms and schemes proposed by those architectures are used in a rather static manner since the involved entities, e.g., the network, sender and receiver, are known before the connection (call) set-up phase. In contrast to these architectures, we propose a general QoS management framework which supports the dynamic choice of a configuration of system components to support the QoS requirements for the user of a specific application. We consider different possible system configurations and select the most appropriate one depending on the desired QoS and the available resources. In this paper we present an overview of this general framework; especially, we concentrate on QoS negotiation and adaptation mechanisms. To show the feasibility of this approach, we designed and implemented a QoS manager for distributed multimedia presentational applications, such as news-on-demand. The negotiation and adaptation mechanisms which are supported by the QoS manager are specializations of the general framework. The proposed framework allows to improve the utilization of system resources, and thus to increase the system availability; it also allows to recover automatically, if this is possible, from QoS degradations. Furthermore, it provides the flexibility to incorporate different resource reservation schemes and scheduling policies, and to accommodate new system component technologies.     

Analysis of bandwidth allocation on end-to-end QoS networks under budget control

This paper considers the problem of bandwidth allocation on communication networks with multiple classes of traffic, where bandwidth is determined under the budget constraint. Due to the limited budget, there is a risk that the network service providers can not assert a 100% guaranteed availability for the stochastic traffic demand at all times. We derive the blocking probabilities of connections as a function of bandwidth, traffic demand and the available number of virtual paths based on the Erlang loss formula for all service classes. A revenue/profit function is studied through the monotonicity and convexity of the blocking probability and expected path occupancy. We present the optimality conditions and develop a solution algorithm for optimal bandwidth of revenue management schemes. The sensitivity analysis and three economic elasticity notions are also proposed to investigate the marginal revenue for a given traffic class by changing bandwidth, traffic demand and the number of virtual paths, respectively. By analysis of those monotone and convex properties, it significantly facilitates the operational process in the efficient design and provision of a core network under the budget constraint.     

Design and performance of speculative flow control for high-radix datacenter interconnect switches

High-radix switches are desirable building blocks for large computer interconnection networks, because they are more suitable to convert chip I/O bandwidth into low latency and low cost than low-radix switches [J. Kim, W.J. Dally, B. Towles, A.K. Gupta, Microarchitecture of a high-radix router, in: Proc. ISCA 2005, Madison, WI, 2005]. Unfortunately, most existing switch architectures do not scale well to a large number of ports, for example, the complexity of the buffered crossbar architecture scales quadratically with the number of ports. Compounded with support for long round-trip times and many virtual channels, the overall buffer requirements limit the feasibility of such switches to modest port counts. Compromising on the buffer sizing leads to a drastic increase in latency and reduction in throughput, as long as traditional credit flow control is employed at the link level. We propose a novel link-level flow control protocol that enables high-performance scalable switches that are based on the increasingly popular buffered crossbar architecture, to scale to higher port counts without sacrificing performance. By combining credited and speculative transmission, this scheme achieves reliable delivery, low latency, and high throughput, even with crosspoint buffers that are significantly smaller than the round-trip time. The proposed scheme substantially reduces message latency and improves throughput of partially buffered crossbar switches loaded with synthetic uniform and non-uniform bursty traffic. Moreover, simulations replaying traces of several typical MPI applications demonstrate communication speedup factors of 2 to 10 times.     

Asynchronous spatial division multiplexing router

Asynchronous quasi-delay-insensitive (QDI) NoCs have several advantages over their clocked counterparts. Virtual channel (VC) is the most utilized flow control method in asynchronous routers but spatial division multiplexing (SDM) achieves better throughput performance for best-effort traffic than VC. A novel asynchronous SDM router architecture is presented. Area and latency models are provided to analyse the network performance of all router architectures including wormhole, virtual channel and SDM. Performance comparisons have been made with different configurations of payload size, communication distance, buffer size, port bandwidth, network size and number of VCs/virtual circuits. Compared with VC, SDM achieves higher throughput with lower area overhead.     

Transaction-Aware Network-on-Chip Resource Reservation

Performance and scalability are critically-important for on-chip interconnect in many-core chip-multiprocessor systems. Packet-switched interconnect fabric, widely viewed as the de facto on-chip data communication backplane in the many-core era, offers high throughput and excellent scalability. However, these benefits come at the price of router latency due to run-time multi-hop data buffering and resource arbitration. The network accounts for a majority of on-chip data transaction latency. In this work, we propose dynamic in-network resource reservation techniques to optimize run-time on-chip data transactions. This idea is motivated by the need to preserve existing abstraction and general-purpose network performance while optimizing for frequently-occurring network events such as data transactions. Experimental studies using multithreaded benchmarks demonstrate that the proposed techniques can reduce on-chip data access latency by 28.4% on average in a 16-node system and 29.2% on average in a 36-node system.     

DHBAM：一种动态层次式带宽分配方法

针对协作式虚拟环境中,协作对象对数据报文的传送有着不同服务要求,提出了一种基于协作层次的动态层次式带宽分配方法。该方法通过带宽预留的方式,能根据协作服务等级和网络链路状况,在主动路由器上对网络带宽进行动态调整,为协作用户提供了高质量的带宽传输保证。仿真结果显示,H、M、L层次的用户由于级别不同而在网络负载变化时能得到适合自身层次的带宽。其分析表明,该方法在系统网络负载变化时能保证不同协作用户之间的数据信息传输的质量要求。     

Preemption-Aware Instantaneous Request Call Routing for Networks With Book-Ahead Reservation

This paper presents a new preemption-aware quality of service (QoS) routing algorithm for instantaneous request (IR) call connections in a QoS-enabled network where resources are shared between IR and book-ahead (BA) call connections. BA reservation, which confirms the availability of resources in advance, is a highly attractive technique for time sensitive applications that require high amount of bandwidth with guaranteed QoS. One of the major concerns for the implementation of BA reservation is the need for preemption of on-going IR calls to accommodate BA calls when resource scarcity arises. Preemption disrupts service continuity of on-going calls which is considered as severely detrimental from users' perceived QoS definition found in recent studies. Existing QoS routing algorithms focus on resource conservation or load balancing as the key objective to attain in addition to guaranteed QoS. No works have yet focused on the preemption problem of on-going IR calls at routing stage in the presence of BA calls. We present a mathematical formulation to compute the preemption probability of an incoming IR call at routing stage based on the current IR and future BA load information. We propose a routing strategy by formulating a link cost function comprising of the calculated preemption probability of the incoming IR call and hop count. Simulation results confirm that QoS routing based on the proposed link cost function significantly outperforms widely recommended shortest path and widest path routing algorithms in terms of IR call preemption and blocking rate. The proposed approach also yields higher network utilization and IR effective throughput.     

Efficient bandwidth utilization in LambdaGrids using pricing incentives

The ability to reserve network bandwidth is a critical factor for the success of high-performance grid applications. Reservation of lightpaths in dynamically switched optical networks facilitates guaranteed bandwidth. However, reservation of bandwidth can often lead to bandwidth fragmentation which significantly reduces system utilization and increases the blocking probability of the network. An interesting approach to mitigating this problem is to induce quasi-flexibility in the user requests. A smart scheduling strategy can then exploit this quasi-flexibility and optimize bandwidth utilization. However, there has to be an incentive for flexibility from the user’s perspective as well. In this paper, we explore how the network service provider (NSP) can influence user flexibility by dynamically engineering pricing incentives. Ultimately, user flexibility will lead to efficient network utilization, reduce the price for the users, and increase the revenue for the NSP.     

LU分解在Godson-Tvl众核体系结构上的半行化研究

随着集成电路工艺的发展,众核体系结构成为人们日益关注的计算平台.LU分解是科学和工程计算中被广泛使用的核心算法之一,尽管在传统的并行体系结构上已有大量的并行化研究工作,但是结合新犁众核体系结构特征的工作还不多.文章从负载均衡、延迟容忍和性能分析模型3个方面系统研究了LU分解在众核体系结构上的并行化问题.该文的贡献在于:首先,针对二维卷帘负载分配方案难以达到良好负载均衡的缺点,提出一种新的"之"字形分配方案,实验表明不经任何优化的情况下性能比前者提高20%,优化后达到了40%;其次,提出了一个性能加速比的分析模型,并用实验定量研究了实测性能加速比和理论值之间的差距,发现在合理利用片上存储优化访存延迟,并恰当选择矩阵分块参数的情况下,实测加速效果能比较接近理论值;通过实验还证明实测性能难以达到理论预测值的两个主要原因:访存带宽有限和片上网络的资源竞争.     

Synchronous bandwidth allocation in FDDI networks

It is well known that an FDDI token ring network provides a guaranteed throughput for synchronous messages and a bounded medium access delay for each node/station. However, this fact alone cannot effectively support many real-time applications that require the timely delivery of each critical message. The reason for this is that the FDDI guarantees a medium access delay bound to nodes, but not to messages themselves. The message-delivery delays may exceed the medium-access delay bound even if a node transmits synchronous messages at a rate not greater than the guaranteed throughput. We solve this problem by developing a synchronous bandwidth allocation (SEA) scheme which calculates the synchronous bandwidth necessary for each application to satisfy its message-delivery delay requirement. The result obtained in this paper is essential for effective use of the FDDI token ring networks in supporting such real-time communication as digital video/audio transmissions, and distributed control/monitoring     

A fair resource allocation protocol for multimedia wireless networks

Wireless networks are expected to support real-time interactive multimedia traffic and must be able, therefore, to provide their users with quality-of-service (QoS) guarantees. Although the QoS provisioning problem arises in wireline networks as well, mobility of hosts and scarcity of bandwidth makes QoS provisioning a challenging task in wireless networks. It has been noticed that multimedia applications can tolerate and gracefully adapt to transient fluctuations in the QoS that they receive from the network. The additional flexibility afforded by the ability of multimedia applications to tolerate and adapt to transient changes in QoS can be exploited by protocol designers to significantly improve the overall performance of wireless systems. This paper presents a fair resource allocation protocol for multimedia wireless networks that uses a combination of bandwidth reservation and bandwidth borrowing to provide network users with QoS in terms of guaranteed bandwidth, call blocking, and call dropping probabilities. Our view of fairness was inspired by the well-known max-min fairness allocation protocol for wireline networks. Simulation results are presented that compare our protocol to similar schemes.     

Affinity-Based Network Interfaces for Efficient Communication on Multicore Architectures

Improving the network interface performance is needed by the demand of applications with high communication requirements (for example, some multimedia, real-time, and high-performance computing applications), and the availability of network links providing multiple gigabits per second bandwidths that could require many processor cycles for communication tasks. Multicore architectures, the current trend in the microprocessor development to cope with the difficulties to further increase clock frequencies and microarchitecture efficiencies, provide new opportunities to exploit the parallelism available in the nodes for designing efficient communication architectures. Nevertheless, although present OS network stacks include multiple threads that make it possible to execute network tasks concurrently in the kernel, the implementations of packet-based or connection-based parallelism are not trivial as they have to take into account issues related with the cost of synchronization in the access to shared resources and the efficient use of caches. Therefore, a common trend in many recent researches on this topic is to assign network interrupts and the corresponding protocol and network application processing to the same core, as with this affinity scheduling it would be possible to reduce the contention for shared resources and the cache misses. In this paper we propose and analyze several configurations to distribute the network interface among the different cores available in the server. These alternatives have been devised according to the affinity of the corresponding communication tasks with the location (proximity to the memories where the different data structures are stored) and characteristics of the processing core. As this approach uses several cores to accelerate the communication path of a given connection, it can be seen as complementary to those that consider several cores to simultaneously process packets belonging to either the same or different connections. Message passing interface (MPI) workloads and dynamic web servers have been considered as applications to evaluate and compare the communication performance of these alternatives. In our experiments, performed by full-system simulation, improvements of up to 35% in the throughput and up to 23% in the latency have been observed in MPI workloads, and up to 100% in the throughput, up to 500% in the response time, and up to 82% in the requests attended per second have been measured in dynamic web servers.