Sorting Reordered Packets with Interrupt Coalescing

Wu, Wenji; DeMar, Phil; Crawford, M.

doi:10.1016/j.comnet.2009.05.012

Cited by 18 publications

(12 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Kandula et al [14] proposed Flare to divide and cache traffic at the granularity of packet bursts, with the observation that if the time between two successive packets is larger than the maximum delay difference between multiple paths, the subsequent packets can be scheduled on any path with no risk of reordering. SPRIC [37] sorts reordered packets in the same block with interrupt coalescing to eliminate or reduce packet reordering at the destination. Other solutions include dynamically adjusting the TCP reordering threshold to avoid improper retransfer or quickly recovering from spurious congestion window shrinks [3,41].…”

Section: Related Workmentioning

confidence: 99%

Real-time video streaming over multipath in multi-hop wireless networks

2010

View full text Add to dashboard Cite

Given the limited wireless link throughput, high loss rate, and varying end-to-end delay, supporting video applications in multi-hop wireless networks becomes a challenging task. Path diversity exploits multiple routes for each session simultaneously, which achieves higher aggregated bandwidth and potentially decreases delay and packet loss. Unfortunately, for TCP-based video streaming, naive load splitting often results in inaccurate estimation of round trip time (RTT) and packet reordering. As a result, it can suffer from significant instability or even throughput reduction, which is also validated by our analysis and simulation in multi-hop wireless networks. To make real-time TCP-based streaming viable over multi-hop wireless networks, we propose a novel cross-layer design with a smart traffic split scheme, namely, multiple path retransmission (MPR). MPR differentiates the original data packets and the retransmitted packets and works with a novel QoS-aware multi-path routing protocol, QAOMDV, to distribute them separately. MPR does not suffer from the RTT underestimation and extra packet reordering, which ensures stable throughput improvement over single-path routing. Through extensive simulations, we further demonstrate that, as compared with state-of-the-art multi-path protocols, our MPR with QAOMDV noticeably enhances the TCP streaming throughput and reduces bandwidth fluctuation, with no obvious impact to fairness.

show abstract

Section: Related Workmentioning

confidence: 99%

Real-time video streaming over multipath in multi-hop wireless networks

2010

View full text Add to dashboard Cite

show abstract

“…There are several studies on network performance optimization in 10Gbps networks [22,24]. However, only a few recent studies have tested highspeed data transfers in a 100Gbps environment.…”

Section: Host Tuning Issuesmentioning

confidence: 99%

Experiences with 100Gbps network applications

Balman

Pouyoul

Yao

et al. 2012

Proceedings of the Fifth International Workshop on Data-Intensive Distributed Computing Date

View full text Add to dashboard Cite

Abstract100Gbps networking has finally arrived, and many research and educational institutions have begun to deploy 100Gbps routers and services. ESnet and Internet2 worked together to make 100Gbps networks available to researchers at the Supercomputing 2011 conference in Seattle Washington. In this paper, we describe two of the first applications to take advantage of this network. We demonstrate a visualization application that enables remotely located scientists to gain insights from large datasets. We also demonstrate climate data movement and analysis over the 100Gbps network. We describe a number of application design issues and host tuning strategies necessary for enabling applications to scale to 100Gbps rates.

show abstract

“…The performance 483 of various reordering-tolerant algorithms on TCP, which distin-484 guishes between normal multipath reordering and loss, has been 485 studied extensively in [28]. We may use lower layer solution on 486 receiving end's network stack as presented in [29]. To address reor-487 dering, on top of our solution, we are considering Linux TCP's adap-488 tive TCP reordering threshold mechanism which is based on the 489 maximum observed reordering length [30].…”

mentioning

confidence: 99%