Here is a list of papers with great artifacts, this list will evolve with time. When you want to discuss a paper, just add it in the list and create a sub-page for this paper with your discussions. Don't hesitate to involve authors in the discussion!

ACM CoNEXT 2016

AmorFi: Amorphous WiFi Networks for High-density Deployments

The static capacity provisioning in traditional WiFi networks (WLANs) cannot cope with the high spatiotemporal traffic variations in high-density venues such as conference centers, stadiums etc. To guarantee reliable performance, venue owners are forced to over-provision their WLANs based on worst-case traffic demand estimations, increasing capital and operational expenses. We propose AmorFi, a radically new way of deploying WLANs to handle peak traffic demands with average-case provisioning. Our key idea is to decouple baseband processing from RF transmission (inspired by the cloud-RAN concept in cellular networks) and introduce software programmability to flexibly allocate WiFi capacity in real time based on varying traffic demands. We implement AmorFi using off-the-shelf WiFi APs over a RF-over-fiber cloud-RAN testbed. Our experiments and simulations demonstrate that the software-defined capacity allocation enabled with AmorFi delivers more than $2x$ throughput than traditional WLANs.

http://dl.acm.org/citation.cfm?id=2999586

[Ramanujan K Sheshadri, Mustafa Y. Arslan, Karthikeyan Sundaresan, Sampath Rangarajan, and Dimitrios Koutsonikolas. 2016. AmorFi: Amorphous WiFi Networks for High-density Deployments. In Proceedings of the 12th International on Conference on emerging Networking EXperiments and Technologies (CoNEXT '16). ACM, New York, NY, USA, 161-175. DOI: https://doi.org/10.1145/2999572.2999586]

ECN or Delay: Lessons Learnt from Analysis of DCQCN and TIMELY

Data center networks, and especially drop-free RoCEv2 networks require efficient congestion control protocols. DCQCN (ECN-based) and TIMELY (delay-based) are two recent proposals for this purpose. In this paper, we analyze DCQCN and TIMELY using fluid models and simulations, for stability, convergence, fairness and flow completion time. We uncover several surprising behaviors of these protocols. For example, we show that DCQCN exhibits non-monotonic stability behavior, and that TIMELY can converge to stable regime with arbitrary unfairness. We propose simple fixes and tuning for ensuring that both protocols converge to and are stable at the fair share point. Finally, using lessons learnt from the analysis, we address the broader question: are there fundamental reasons to prefer either ECN or delay for end-to-end congestion control in data center networks? We argue that ECN is a better congestion signal, due to the way modern switches mark packets, and due to a fundamental limitation of end-to-end delay-based protocols, that we derive.

http://dl.acm.org/citation.cfm?id=2999593

[Yibo Zhu, Monia Ghobadi, Vishal Misra, and Jitendra Padhye. 2016. ECN or Delay: Lessons Learnt from Analysis of DCQCN and TIMELY. In Proceedings of the 12th International on Conference on emerging Networking EXperiments and Technologies (CoNEXT '16). ACM, New York, NY, USA, 313-327. DOI: https://doi.org/10.1145/2999572.2999593]

HyPer4: Using P4 to Virtualize the Programmable Data Plane

Through virtualization, single physical data planes can logically support multiple networking contexts. We propose HyPer4 as a portable virtualization solution. HyPer4 provides a general purpose program, written in the P4 dataplane programming language, that may be dynamically configured to adopt behavior that is functionally equivalent to other P4 programs. HyPer4 extends, through software, the following features to diverse P4-capable devices: the ability to logically store multiple programs and either run them in parallel (network slicing) or as hot-swappable snapshots; and virtual networking between programs (supporting program composition or multi-tenant service interaction). HyPer4 permits modifying the set of programs, as well as the virtual network connecting them, at runtime, without disrupting currently active programs. We show that realistic ASICs-based hardware would be capable of running HyPer4 today.

http://dl.acm.org/citation.cfm?id=2999607

[David Hancock and Jacobus van der Merwe. 2016. HyPer4: Using P4 to Virtualize the Programmable Data Plane. In Proceedings of the 12th International on Conference on emerging Networking EXperiments and Technologies (CoNEXT '16). ACM, New York, NY, USA, 35-49. DOI: https://doi.org/10.1145/2999572.2999607]

LossRadar: Fast Detection of Lost Packets in Data Center Networks

Packet losses are common in data center networks, may be caused by a variety of reasons (e.g., congestion, blackhole), and have significant impacts on application performance and network operations. Thus, it is important to provide fast detection of packet losses independent of their root causes. We also need to capture both the locations and packet header information of the lost packets to help diagnose and mitigate these losses. Unfortunately, existing monitoring tools that are generic in capturing all types of network events often fall short in capturing losses fast with enough details and low overhead. Due to the importance of loss in data centers, we propose a specific monitoring system designed for loss detection. We propose LossRadar, a system that can capture individual lost packets and their detailed information in the entire network on a fine time scale. Our extensive evaluation on prototypes and simulations demonstrates that LossRadar is easy to implement in hardware switches, achieves low memory and bandwidth overhead, while providing detailed information about individual lost packets. We also build a loss analysis tool that demonstrates the usefulness of LossRadar with a few example applications.

http://dl.acm.org/citation.cfm?id=2999609

[Yuliang Li, Rui Miao, Changhoon Kim, and Minlan Yu. 2016. LossRadar: Fast Detection of Lost Packets in Data Center Networks. In Proceedings of the 12th International on Conference on emerging Networking EXperiments and Technologies (CoNEXT '16). ACM, New York, NY, USA, 481-495. DOI: https://doi.org/10.1145/2999572.2999609]

PI2: A Linearized AQM for both Classic and Scalable TCP

This paper concerns the use of Active Queue Management (AQM) to reduce queuing delay. It offers insight into why it has proved hard for a Proportional Integral (PI) controller to remain both responsive and stable while controlling `Classic' TCP flows, such as TCP Reno and Cubic. Due to their non-linearity, the controller's adjustments have to be smaller when the target drop probability is lower. The PI Enhanced (PIE) algorithm attempts to solve this problem by scaling down the adjustments of the controller using a look-up table. Instead, we control an internal variable that is by definition linearly proportional to the load, then post-process it into the required Classic drop probability---in fact we show that the output simply needs to be squared. This allows tighter control, giving responsiveness and stability better or no worse than PIE achieves, but without all its corrective heuristics.

Additionally, with suitable packet classification, it becomes simple to extend this PI2 AQM to support coexistence between Classic and Scalable congestion controls in the public Internet. Unlike a Classic congestion control, a Scalable congestion control ensures sufficient feedback at any flow rate, an example being Data Centre TCP (DCTCP). A Scalable control is linear, so we can use the internal variable directly without any squaring, by omitting the post-processing stage.

We implemented this PI2 AQM as a Linux qdisc to extensively test our claims using Classic and Scalable TCPs.

http://dl.acm.org/citation.cfm?id=2999578

[Koen De Schepper, Olga Bondarenko, Ing-Jyh Tsang, and Bob Briscoe. 2016. PI2: A Linearized AQM for both Classic and Scalable TCP. In Proceedings of the 12th International on Conference on emerging Networking EXperiments and Technologies (CoNEXT '16). ACM, New York, NY, USA, 105-119. DOI: https://doi.org/10.1145/2999572.2999578]

RT-OPEX: Flexible Scheduling for Cloud-RAN Processing

It is cost-effective to process wireless frames on general purpose processors (GPPs) in place of dedicated hardware. Wireless operators are decoupling signal processing from basestations and implementing it in a cloud of compute resources, also known as a cloud-RAN (C-RAN). A C-RAN must meet the deadlines of processing wireless frames; for example, 3ms to transport, decode and respond to an LTE uplink frame. The design of baseband processing on these platforms is thus a major challenge for which various processing and real-time scheduling techniques have been proposed. In this paper, we implement a medium-scale C-RAN-type platform and conduct an in-depth analysis of its real-time performance. We find that the commonly used (e.g., partitioned) scheduling techniques for wireless frame processing are inefficient as they either over-provision resources or suffer from deadline misses. This inefficiency stems from the large variations in processing times due to fluctuations in wireless traffic. We present a new framework called RTOPEX, that leverages these variations and proposes a flexible approach for scheduling. RT-OPEX dynamically migrates parallelizable tasks to idle compute resources at runtime, reducing processing times and hence deadline misses at no additional cost. We implement and evaluate RT-OPEX on a commodity GPP platform using realistic cellular workload traces. Our results show that RT-OPEX achieves an order-of-magnitude improvement over existing C-RAN schedulers in meeting frame processing deadlines.

http://dl.acm.org/citation.cfm?id=2999591

[Krishna C. Garikipati, Kassem Fawaz, and Kang G. Shin. 2016. RT-OPEX: Flexible Scheduling for Cloud-RAN Processing. In Proceedings of the 12th International on Conference on emerging Networking EXperiments and Technologies (CoNEXT '16). ACM, New York, NY, USA, 267-280. DOI: https://doi.org/10.1145/2999572.2999591]

Xpander: Towards Optimal-Performance Datacenters

Despite extensive efforts to meet ever-growing demands, today's datacenters often exhibit far-from-optimal performance in terms of network utilization, resiliency to failures, cost efficiency, incremental expandability, and more. Consequently, many novel architectures for high-performance datacenters have been proposed. We show that the benefits of state-of-the-art proposals are, in fact, derived from the fact that they are (implicitly) utilizing "expander graphs" (aka expanders) as their network topologies, thus unveiling a unifying theme of these proposals. We observe, however, that these proposals are not optimal with respect to performance, do not scale, or suffer from seemingly insurmountable deployment challenges. We leverage these insights to present Xpander, a novel datacenter architecture that achieves near-optimal performance and provides a tangible alternative to existing datacenter designs. Xpander's design turns ideas from the rich graph-theoretic literature on constructing optimal expanders into an operational reality. We evaluate Xpander via theoretical analyses, extensive simulations, experiments with a network emulator, and an implementation on an SDN-capable network testbed. Our results demonstrate that Xpander significantly outperforms both traditional and proposed datacenter designs. We discuss challenges to real-world deployment and explain how these can be resolved.

http://dl.acm.org/citation.cfm?id=2999580&dl=ACM&coll=DL&CFID=720345214&CFTOKEN=87339349

[Asaf Valadarsky, Gal Shahaf, Michael Dinitz, and Michael Schapira. 2016. Xpander: Towards Optimal-Performance Datacenters. In Proceedings of the 12th International on Conference on emerging Networking EXperiments and Technologies (CoNEXT '16). ACM, New York, NY, USA, 205-219. DOI: https://doi.org/10.1145/2999572.2999580]

Tags:
Created by Damien Saucez on 2017/01/24 15:48