Buffer sizing in all-optical packet switches

Beheshti, Neda; Ganjali, Yashar; Rajaduray, R.; Blumenthal, Daniel J.; McKeown, Nick

doi:10.1109/ofc.2006.215714

Cited by 40 publications

(31 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The goal of buffer sizing is to find out how small we can make Internet router buffers without any degradation in network performance. A plethora of recent work emerged to reduce buffer sizes [1,2,3,4,5] and to understand the relationships between buffer sizing and other parameters of the network [8,9,10,11,12,13,14], such as, throughput, delay, loss, stability [7,6], and the impacts of various traffic conditions [8,15].…”

Section: Motivation and Introductionmentioning

confidence: 99%

The Effects of Fairness in Buffer Sizing

Wang

Ganjali

2007

NETWORKING 2007. Ad Hoc and Sensor Networks, Wireless Networks, Next Generation Internet

View full text Add to dashboard Cite

Abstract. Buffer sizing in Internet routers is a fundamental problem that has major consequences in the design, implementation, and economy of the routers, as well as on the performance observed by the end users. Recently, there have been some seemingly contradictory results on buffer sizing. On the one hand, Appenzeller et al. show that as a direct consequence of desynchronization of flows in the core of the Internet, buffer sizes in core routers can be significantly reduced without any major degradation in network performance. On the other hand, Raina and Wischik show that such reduction in buffer sizing comes at the cost of synchronization and thus instability in the network.This work unifies these results by studying the effects of fairness in buffer sizing. We show that the main difference arises from the implicit assumption of fairness in packet dropping in the latter result. We demonstrate that desynchronization among flows observed by Appenzeller et al. is caused by unfair packet dropping when a combination of TCP-Reno and the drop-tail queue management is used. We also show that bringing fairness in packet dropping will introduce synchronization among flows, and will make the system unstable as predicted by Raina and Wischik.Our analysis suggests that there is an intrinsic trade-off between fairness in packet drops and desynchronization among TCP-Reno flows when routers use the drop-tail queue management. Achieving fairness, desynchronization, small buffer size, and 100% link utilization at the same time is desirable and feasible yet challenging. The studies in this paper provide insights for further explorations in reaching this goal.

show abstract

Section: Motivation and Introductionmentioning

confidence: 99%

The Effects of Fairness in Buffer Sizing

Wang

Ganjali

2007

NETWORKING 2007. Ad Hoc and Sensor Networks, Wireless Networks, Next Generation Internet

View full text Add to dashboard Cite

show abstract

“…Testbeds for networking can be shared among many researchers [9,10,11,12] or specific to one project [13,14,15]. Though realistic, they cost money to build and keep running, have practical resource limits (especially before paper deadlines), and may lack the flexibility to support experiments with custom topologies or custom forwarding behavior.…”

Section: Goalsmentioning

confidence: 99%

Reproducible network experiments using container-based emulation

Handigol

Heller

Jeyakumar

et al. 2012

Proceedings of the 8th International Conference on Emerging Networking Experiments and Technologies

402

204

View full text Add to dashboard Cite

In an ideal world, all research papers would be runnable: simply click to replicate all results, using the same setup as the authors. One approach to enable runnable network systems papers is Container-Based Emulation (CBE), where an environment of virtual hosts, switches, and links runs on a modern multicore server, using real application and kernel code with software-emulated network elements. CBE combines many of the best features of software simulators and hardware testbeds, but its performance fidelity is unproven.In this paper, we put CBE to the test, using our prototype, Mininet-HiFi, to reproduce key results from published network experiments such as DCTCP, Hedera, and router buffer sizing. We report lessons learned from a graduate networking class at Stanford, where 37 students used our platform to replicate 16 published results of their own choosing. Our experiences suggest that CBE makes research results easier to reproduce and build upon.

show abstract

“…While some earlier works such as [3], [11] have applied buffer sizing principles to optical switches, they focus entirely on TCP traffic performance, and completely ignore the performance implications for real-time traffic. Though it is argued that 90-95% of today's Internet traffic is carried over (closed-loop) TCP, there is evidence to indicate that there is a growing demand for (open-loop) real-time traffic, driven by the rising popularity of audio/video, online gaming, VoIP, and IPTV applications.…”

Section: A Buffer Sizing In Optical Networkmentioning

confidence: 99%

Considerations for Sizing Buffers in Optical Packet Switched Networks

2009

View full text Add to dashboard Cite

Abstract-Optical packet switches of the foreseeable future are expected to have severely limited buffering capability, since storage of optical signals remains a difficult and expensive operation. Our observations in simulation of TCP and realtime traffic in networks with such small buffers have revealed regions of anomalous performance in which losses for real-time traffic become higher as buffers get larger. The detrimental impact of larger optical buffers is studied in this paper and three new contributions are made. First, we develop a Markov chain model that allows analytical computation of loss. Our model validates observations from simulation, and opens the doors to an analytical understanding of how various factors affect the anomaly. Second, we study the anomaly under realistic traffic mixes containing persistent and non-persistent TCP flows, and show that the traffic mix does not significantly alter the anomaly. Third, we show that larger diversity in packet size between TCP and real-time traffic increases the severity of the anomaly, and is an important consideration when sizing optical switch buffers, particularly since real-time and TCP ACK packets are significantly smaller than the TCP data packets. Our study informs switch manufacturers and network operators of factors to consider when selecting optical buffer sizes in order to achieve desired performance balance between TCP and real-time traffic.

show abstract

Buffer sizing in all-optical packet switches

Abstract: Packet-switched routers need buffers during times of congestion. We show that a combined input-output queued router needs no more buffering than an output queued router. Using simulations, we show that 10-20 packet buffers are enough.

Cited by 40 publications

References 6 publications

The Effects of Fairness in Buffer Sizing

The Effects of Fairness in Buffer Sizing

Reproducible network experiments using container-based emulation

Considerations for Sizing Buffers in Optical Packet Switched Networks

Contact Info

Product

Resources

About