Achieving 100% throughput in an input-queued switch

McKeown, Nick; Mekkittikul, A.; Anantharam, Venkat; Walrand, Jean

doi:10.1109/26.780463

Cited by 684 publications

(98 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…matchings) may be simultaneously connected because of compatibility constraints, see for instance McKeown et al [13], [14]. The book of Meyn [15] contains extensive background material on MaxWeight policies.…”

Section: Introductionmentioning

confidence: 99%

Spatial inefficiency of MaxWeight scheduling

Ven

Borst

Lü

2011

2011 International Symposium of Modeling and Optimization of Mobile, Ad Hoc, and Wireless Networks

View full text Add to dashboard Cite

Abstract-MaxWeight scheduling has gained enormous popularity as a powerful paradigm for achieving queue stability and maximum throughput in a wide variety of scenarios. The maximum-stability guarantees however rely on the fundamental premise that the system consists of a fixed set of flows with stationary ergodic traffic processes. In the present paper we examine networks where the population of active flows varies over time, as flows eventually end while new flows occasionally start. We show that MaxWeight policies may fail to provide maximum stability due to persistent inefficient spatial reuse. The intuitive explanation is that these policies tend to serve flows with large backlogs, even when the resulting spatial reuse is not particularly efficient, and fail to exploit maximum spatial reuse patterns involving flows with smaller backlogs. These results indicate that instability of MaxWeight scheduling can occur due to spatial inefficiency in networks with fixed transmission rates, which is fundamentally different from the inability to fully exploit timevarying rates shown in prior work. We discuss how the potential instability effects can be countered by spatial traffic aggregation, and describe some of the associated challenges and performance trade-offs.

show abstract

Section: Introductionmentioning

confidence: 99%

Spatial inefficiency of MaxWeight scheduling

Ven

Borst

Lü

2011

2011 International Symposium of Modeling and Optimization of Mobile, Ad Hoc, and Wireless Networks

View full text Add to dashboard Cite

show abstract

“…We are ready to provide an upper bound on the first term in the LHS of (15). Noting the fact that the amount of unused service in one time-slot at each link cannot be greater than the maximum channel rate , i.e., , we have (17) where the last step utilizes inequality (16). Next, we focus on the second term in the LHS of (15).…”

Section: B Proof Of Main Resultsmentioning

confidence: 99%

“…Yet, both the time-varying nature of wireless channels and the scheduling policy significantly affect the regularity of the received data of each mobile user. The traditional scheduling policies aiming to maximize the system throughput (e.g., [10], [17], and [22]) or provide various fairness guarantees (e.g., [3], [16], [18], and [21]) at the base station side do not take users' experience into account and thus lead to the high irregularity of the received data of mobile users.…”

mentioning

confidence: 99%

Wireless Scheduling Design for Optimizing Both Service Regularity and Mean Delay in Heavy-Traffic Regimes

Li²,

Eryılmaz

2016

IEEE/ACM Trans. Networking

View full text Add to dashboard Cite

We consider the design of throughput-optimal scheduling policies in multihop wireless networks that also possess good mean delay performance and provide regular service for all links-critical metrics for real-time applications. To that end, we study a parametric class of maximum-weight-type scheduling policies, called Regular Service Guarantee (RSG) Algorithm, where each link weight consists of its own queue length and a counter that tracks the time since the last service, namely Time-Since-Last-Service (TSLS). The RSG Algorithm not only is throughput-optimal, but also achieves a tradeoff between the service regularity performance and the mean delay, i.e., the service regularity performance of the RSG Algorithm improves at the cost of increasing mean delay. This motivates us to investigate whether satisfactory service regularity and low mean-delay can be simultaneously achieved by the RSG Algorithm by carefully selecting its design parameter. To that end, we perform a novel Lyapunov-drift-based analysis of the steady-state behavior of the stochastic network. Our analysis reveals that the RSG Algorithm can minimize the total mean queue length to establish mean delay optimality under heavily loaded conditions as long as the design parameter weighting for the TSLS scales no faster than the order of , where measures the closeness of the network load to the boundary of the capacity region. To the best of our knowledge, this is the first work that provides regular service to all links while also achieving heavy-traffic optimality in mean queue lengths.Index Terms-Heavy-traffic analysis, mean delay, service regularity, throughput, wireless scheduling. 1063-6692

show abstract

“…In a simple switching network, a router or switch is designed to operate in a chassis or rack, and can employ a centralized scheduler to compute a match between ingress and egress ports in each transmission time-slot [11]. In a more generalized network, a centralized scheduler can perform path selection and bandwidth allocation for each traffic flow [12].…”

Section: A Related Workmentioning

confidence: 99%

Joint Static and Dynamic Traffic Scheduling in Data Center Networks

Cao

Kodialam²,

Lakshman³

2016

IEEE/ACM Trans. Networking

View full text Add to dashboard Cite

The advent and continued growth of large data centers has led to much interest in switch architectures that can economically meet the high capacities needed for interconnecting the thousands of servers in these data centers. Various multilayer architectures employing thousands of switches have been proposed in the literature. We make use of the observation that the traffic in a data center is a mixture of relatively static and rapidly fluctuating components, and develop a combined scheduler for both these components using a generalization of the load-balanced scheduler. The presence of the known static component introduces asymmetries in the ingress-egress capacities, which preclude the use of a load-balanced scheduler as is. We generalize the load-balanced scheduler and also incorporate an opportunistic scheduler that sends traffic on a direct path when feasible to enhance the overall switch throughput. Our evaluations show that this scheduler works very well despite avoiding the use of a central scheduler for making packet-by-packet scheduling decisions.Index Terms-Data center network, load-balanced scheduler, traffic scheduling. 1063-6692

show abstract

Achieving 100% throughput in an input-queued switch

Cited by 684 publications

References 18 publications

Spatial inefficiency of MaxWeight scheduling

Spatial inefficiency of MaxWeight scheduling

Wireless Scheduling Design for Optimizing Both Service Regularity and Mean Delay in Heavy-Traffic Regimes

Joint Static and Dynamic Traffic Scheduling in Data Center Networks

Contact Info

Product

Resources

About