Authors: Zhiqiang He (USTC, China; Huawei, China), Dongyang Wang (Huawei, China), Binzhang Fu (Huawei, China), Kun Tan (Huawei, China), Bei Hua (USTC, China), Zhi-Li Zhang(University of Minnesota, USA), Kai Zheng (Huawei, China)
Abstract: RDMA communication in virtual private cloud (VPC) networks is still a challenging job due to the difficulty in fulfilling all virtualization requirements without sacrificing RDMA communication performance. To address this problem, this paper proposes a software-defined solution, namely, MasQ, which is short for “queue masquerade”. The core insight of MasQ is that all RDMA communications should associate with at least one queue pair (QP). Thus, the requirements of virtualization, such as network isolation and the application of security rules, can be easily fulfilled if QP’s behavior is properly defined. In particular, MasQ exploits the virtio-based paravirtualization technique to realize the control path. Moreover, to avoid performance overhead, MasQ leaves all data path operations, such as sending and receiving, to the hardware. We have implemented MasQ in the OpenFabrics Enterprise Distribution (OFED) framework and proved its scalability and performance efficiency by evaluating it against typical applications. The results demonstrate that MasQ achieves almost the same performance as bare-metal RDMA for data communication.
Speaker Bio: Zhiqiang He is now pursuing his PhD degree at USTC. His research interests include RDMA, network virtualization, and cloud networking. His advisor is Prof. [Bei Hua](http://staff.ustc.edu.cn/~bhua).
Authors: Shuihai Hu, Wei Bai, Gaoxiong Zeng, Zilong Wang, Baochen Qiao, Kai Chen, Kun Tan, Yi Wang
Abstract: As datacenter network bandwidth keeps growing, proactive transport becomes attractive, where bandwidth is proactively allocated as “credits” to senders who then can send “scheduled packets” at a right rate to ensure high link utilization, low latency, and zero packet loss. While promising, a fundamental challenge is that proactive transport requires at least one-RTT for credits to be computed and delivered. In this paper, we show such one-RTT “pre-credit” phase could carry a substantial amount of flows at high link-speeds, but none of existing proactive solutions treats it appropriately. We present Aeolus, a solution focusing on “pre-credit” packet transmission as a building block for proactive transports. Aeolus contains unconventional design principles such as scheduled-packet-first (SPF) that de-prioritizes the first-RTT packets, instead of prioritizing them as prior work. It further exploits the preserved, deterministic nature of proactive transport as a means to recover lost first-RTT packets efficiently. We have integrated Aeolus into ExpressPass, NDP and Homa, and shown, through both implementation and simulations, that the Aeolus-enhanced solutions deliver significant performance or deployability advantages. For example, it improves the average FCT of ExpressPass by 56%, cuts the tail FCT of Homa by 20×, while achieving similar performance as NDP without switch modifications.
Speaker Bio: Shuihai Hu is currently the Chief Scientist at Clustar. He received his PhD degree in computer science from HKUST in 2019. Before that, He received his B.S. degree in computer science from USTC in 2013. Shuihai’s current interests are mainly about datacenter networks and machine learning systems (with a special focus on federated machine learning).
Abstract: Network telemetry is essential for administrators to monitor massive data traffic in a network-wide manner. Existing telemetry solutions often face the dilemma between resource efficiency (i.e., low CPU, memory, and bandwidth overhead) and full accuracy (i.e., error-free and holistic measurement). We break this dilemma via a network-wide architectural design OmniMon, which simultaneously achieves resource efficiency and full accuracy in flow-level telemetry for large-scale data centers. OmniMon carefully coordinates the collaboration among different types of entities in the whole network to execute telemetry operations, such that the resource constraints of each entity are satisfied without compromising full accuracy. It further addresses consistency in network-wide epoch synchronization and accountability in error-free packet loss inference. We prototype OmniMon in DPDK and P4. Testbed experiments on commodity servers and Tofino switches demonstrate the effectiveness of OmniMon over state-of-the-art telemetry designs.
Speaker Bio: Qun Huang is an Assistant Professor (Tenure-Track) at Department of Computer Science and Technology, Peking University. His research focuses on network measurement now.
Abstract: The shared nature of the wireless medium induces contention between data transport and backward signaling, such as acknowledgement. The current way of TCP acknowledgment induces control overhead which is counter-productive for TCP performance especially in wireless local area network (WLAN) scenarios. In this paper, we present a new acknowledgement called TACK (“Tame ACK”), as well as its TCP implementation TCP-TACK. TCP-TACK works on top of commodity WLAN, delivering high wireless transport goodput with minimal control overhead in the form of ACKs, without any hardware modification. To minimize ACK frequency, TACK abandons the legacy received-packet-driven ACK. Instead, it balances byte-counting ACK and periodic ACK so as to achieve a controlled ACK frequency. Evaluation results show that TCP-TACK achieves significant advantages over legacy TCP in WLAN scenarios due to less contention between data packets and ACKs. Specifically, TCP-TACK reduces over 90% of ACKs and also obtains an improvement of ∼28% on goodput. We further find it performs equally well as high-speed TCP variants in wide area network (WAN) scenarios, this is attributed to the advancements of the TACK-based protocol design in loss recovery, round-trip timing, and send rate control.
Speaker Bio: Tong Li received his B.S. degree from Wuhan University in 2012 and his Ph.D. degree from Tsinghua University in 2017. He is currently a senior researcher in the Computer Network and Protocol Lab at Huawei. During the last three years, his research focuses on transport protocol measurement, design, and implementation in both wired and wireless networks, two of which are published in SIGCOMM.
Abstract:This talk presents Hoyan– the first reported large scale deployment of configuration verification in a global-scale wide area network (WAN). Hoyan has been running in production for more than two years and is currently used for all critical configuration auditing and updates on the WAN. We highlight our innovative designs and real-life experience to make Hoyan accurate and scalable in practice. For accuracy under the inconsistencies of devices’ vendor-specific behaviors (VSBs), Hoyan continuously discovers the flaws in device behavior models, thus aiding the operators in fixing the models. For scalability to verify our global WAN, Hoyan introduces a “global-simulation & local formal-modeling” strategy to model uncertainties in small scales and perform aggressive pruning of possibilities during the protocol simulations. Hoyan achieves near-100% verification accuracy after it detected and fixed O(10) VSBs on our WAN. Hoyan has prevented many potential service failures resulting from misconfiguration and reduced the failure rate of updates of our WAN by more than half in 2019.
Speaker Bio: Hongqiang "Harry" Liu leads a network research team in Alibaba Cloud. His research focuses on high speed and programmable networking and intent-based networking. Before joining Alibaba, Dr. Liu was a Senior Researcher in Microsoft Research Redmond Lab, where he created the well-known large-scale network emulator, CrystalNet [SOSP’17], which has been widely used in industry. He received his Ph.D. degree from the Department of Computer Science at Yale University in 2014 and his Master's and Bachelor's degrees from the Department of Electronic Engineering, Tsinghua University, China. His research interest lies on many fields of networking and cloud computing, including high performance networking (e.g. RDMA), programmable data plane networking (P4), network verification and testing, virtual networking, container networking, and edge computing. Dr. Liu has published 20 papers in top-tier academic conferences, such as ACM SIGCOMM, ACM SOSP, USENIX NSDI. He is the recipient of the prestigious ACM SIGCOMM Doctoral Dissertation Award - Honorable Mention in 2015.
Authors: Chongrong Fang (Zhejiang University),Haoyu Liu (Zhejiang University),Mao Miao (Alibaba Group),Jie Ye (Alibaba Group),Lei Wang (Alibaba Group),Wansheng Zhang (Alibaba Group),Daxiang Kang (Alibaba Group),Biao Lyv (Alibaba Group),Peng Cheng (Zhejiang University),Jiming Chen (Zhejiang University)
Abstract:With the characteristics of remarkable flexibility and scalability, overlay network has been the key technology for the cloud computing. Compared with underlay network, larger scale and complexity of overlay network have outpaced traditional naive tools for monitoring and debugging. Existing methods such as traceroute or other cutting-edge techniques, focus on diagnosing physical network in large datacenters, and are insufficient for debugging global-wide overlay network. It poses significant challenges to inspect overlay network due to its heavier data traffic, larger topology complexity and higher demand for quick response. We present VTrace, a non-intrusive and responsive network traceback analytic system which tracks arbitrary target packet over cloud-scale overlay network. For the sake of convenience, VTrace reports and records target packets automatically with a set of "matching, coloring and logging" rules maintained in virtual forwarding devices. Log Agents in different regions are applied to store network-wide logs, while a distributed stream processing platform is utilized for analyzing virtual flows timely. Experiments are conducted in a test environment showing that VTrace brings negligible effect on production traffic and takes averagely 0.54ms to reconstruct one flow when performing 800 concurrent tasks.
Speaker Bio: Biao Lyu receives his bachelor degrees in computer science from both ZJU and SFU. Since then he has been working on network virtualization and cloud networking in Microsoft Windows team and Alibaba Cloud Network Product team. Now he is leading the smart network operation and analytics team Qitian in Alibaba Cloud, focused on operating Alibaba cloud network in a data-driven pattern using big data and machine learning technology.
Abstract: Current UHF RFID systems suffer from two long-standing problems: 1) miss-reading non-line-of-sight or mis-oriented tags and 2) cross-reading undesired, distant tags due to multi-path reflections. We propose a novel system, NFC+, to overcome the fundamental challenges. NFC+ is a magnetic field reader, which can inventory standard NFC tagged objects with a reasonably long range and arbitrary orientation. NFC+ achieves this by leveraging physical and algorithmic techniques based on magnetic resonance engineering. We build a prototype of NFC+ and conduct extensive evaluations in a logistic network. Comparing to UHF RFID, we find that NFC+ can reduce the miss-reading rate from 23% to 0.03%, and cross-reading rate from 42% to 0, for randomly oriented objects. NFC+ demonstrates high robustness for RFID unfriendly media (e.g., water bottles and metal cans). It can reliably read commercial NFC tags at a distance of up to 3 meters which, for the first time, enables NFC to be directly applied to practical logistics network applications. To our knowledge, NFC+ represents the first system to solve the cross-reading and miss-reading problem that has plagued RFID for decades in realistic deployment. We believe that NFC+ paves the way towards deploying RFID at scale in logistics networks.
Speaker Bio: Yunfei Ma is currently a senior research engineer in the Network Research team at Alibaba Group US and the XG Lab in the DAMO Academy. Before joining Alibaba, He was a postdoctoral scholar at MIT Media Lab. His research interests include wireless networking systems, AIoT and multimedia transport protocols. He received Ph.D. in Electrical and Computer Engineering from Cornell University and B.S. degree from USTC. He holds 9 granted US patents and has more than 10 publications on top CS conferences including SIGCOMM, MOBICOM and NSDI. His research has been covered by media outlets including BBC, The Verge, MIT Technology Review and IEEE Spectrum. He has served on the TPC of ACM CoNEXT 2018 and IEEE INFOCOM 2020/2019/2018.
Authors:Zili Meng, Minhu Wang, Jiasong Bai, Mingwei Xu (Tsinghua University), Hongzi Mao (MIT), Hongxin Hu (Clemson University)
Abstract:While many deep learning (DL)-based networking systems have demonstrated superior performance, the underlying Deep Neural Networks (DNNs) remain blackboxes and stay uninterpretable for network operators. The lack of interpretability makes DL-based networking systems prohibitive to deploy in practice. In this paper, we propose Metis, a framework that provides interpretability for two general categories of networking problems spanning local and global control. Accordingly, Metis introduces two different interpretation methods based on decision tree and hypergraph, where it converts DNN policies to interpretable rule-based controllers and highlight critical components based on analysis over hypergraph. We evaluate Metis over two categories of state-of-the-art DL-based networking systems and show that Metis provides human-readable interpretations while preserving nearly no degradation in performance. We further present four concrete use cases of Metis, showcasing how Metis helps network operators to design, debug, deploy, and ad-hoc adjust DL-based networking systems.
Speaker Bio: Zili Meng received his B.Eng. degree from Tsinghua University in 2019, where he is now pursuing his PhD degree. His research interests include the intersection of machine learning and networking systems. His advisor is Prof. Mingwei Xu.
Abstract: Network performance anomalies (NPAs), e.g. long-tailed latency, bandwidth decline, etc., are increasingly crucial to cloud providers as applications are getting more sensitive to performance. The fundamental difficulty to quickly mitigate NPAs lies in the limitations of state-of-the-art network monitoring solutions — coarse-grained counters, active probing, or packet telemetry either cannot provide enough insights on flows or incur too much overhead. This paper presents NetSeer, a flow event telemetry (FET) monitor which aims to discover and record all performance-critical data plane events, e.g. packet drops, congestion, path change, and packet pause. NetSeer is efficiently realized on the programmable data plane. It has a high coverage on flow events including inter-switch packet drop/corruption which is critical but also challenging to retrieve the original flow information, with novel intra- and inter-switch event detection algorithms running on data plane; NetSeer also achieves high scalability and accuracy with innovative designs of event aggregation, information compression, and message batching that mainly run on data plane, using switch CPU as complement. NetSeer has been implemented on commodity programmable switches and NICs. With real case studies and extensive experiments, we show NetSeer can reduce NPA mitigation time by 61%–99% with only 0.01% overhead of monitoring traffic.
Speaker Bio:Chen Sun is a network engineer in the Network Research Team at Alibaba Group. He received his Ph.D degree from the Department of Computer Science and Technology in Tsinghua University in 2019. His research interests include SDN/NFV, programmable networks, and network monitoring. He has published over 10 papers on top conferences and journals.
Authors: Dongzhu Xu, Anfu Zhou, Xinyu Zhang, Guixian Wang, Xi Liu, Congkai An, Yiming Shi, Liang Liu, Huadong Ma
Abstract:5G, as a monumental shift in cellular communication technology, holds tremendous potential for spurring innovations across many vertical industries, with its promised multi-Gbps speed, sub-10 ms low latency, and massive connectivity. On the other hand, as 5G has been deployed for only a few months, it is unclear how well and whether 5G can eventually meet its prospects. In this paper, we demystify operational 5G networks through a first-of-its-kind cross-layer measurement study. Our measurement focuses on four major perspectives: (i) Physical layer signal quality, coverage and hand-off performance; (ii) End-to-end throughput and latency; (iii) Quality of experience of 5G’s niche applications (e.g., 4K/5.7K panoramic video telephony); (iv) Energy consumption on smartphones. The results reveal that the 5G link itself can approach Gbps throughput, but legacy TCP leads to surprisingly low capacity utilization (< 32%), latency remains too high to support tactile applications and power consumption escalates to 2 - 3× over 4G. Our analysis suggests that the wireline paths, upper-layer protocols, computing and radio hardware architecture need to co- evolve with 5G to form an ecosystem, in order to fully unleash its potential.
Speaker Bio:He is a first-year Ph.D. student from Beijing University of Posts and Telecommunications (BUPT), majoring in Computer Science and Technology. His research interests include network measurement, Open 5G, and MEC.
Authors: Yue Wu, Purui Wang, Kenuo Xu, Lilei Feng, Chenren Xu (Peking University)
Abstract: Visible light backscatter communication (VLBC) presents an emerging low power IoT connectivity solution with spatial reuse and interference immunity advantages over RF-based (backscatter) technologies. State-of-the-art VLBC systems employ COTS LCD shutter as optical modulator, whose slow response fundamentally throttles its data rate to sub-Kbps, and limits its deployment at scale for use cases where higher rate and/or low latency is a necessity. We design and implement RetroTurbo, a VLBC system dedicated for turboboosting data rate. At the heart of RetroTurbo design is a pair of novel modulation schemes, namely delayed superimposition modulation (DSM) and polarization-based QAM (PQAM), to push the rate limit by strategically coordinating the state of a liquid crystal modulator (LCM) pixel array in time and polarization domains. Specifically, DSM ensures we fully exploit the available SNR for high order modulation in the LCM-imposed nonlinear channel; PQAM is based on polarized light communication that creates a QAM design in polarization domain with flexible angular misalignment between two ends. A real-time near-optimal demodulation algorithm is designed to ensure system's robustness to heterogeneous signal distortion. Based on our prototyped system, RetroTurbo demonstrates 32x and 128x rate gain via experiments and emulation respectively in practical real-world indoor setting.
Speaker Bio: Yue Wu is a senior undergraduate student at Peking University and will pursue his Ph.D. degree at Yale University. His research interest concentrates on computer systems, specifically, wireless network systems and embedded systems. He has great passion in developing software-hardware orchestrated solutions for real-world problems. During the last three years, he worked on visible light backscatter communication system and wireless charging system which was published in SIGCOMM, MobiCom, etc. Besides, he explored massive-MIMO system and evolvability in operating system which broadened his horizon. His advisor is Prof. Chenren Xu.
Authors: Jiaqi Gao, Ennan Zhai, Hongqiang Harry Liu, Rui Miao, Yu Zhou, Bingchuan Tian, Chen Sun, Dennis Cai, Ming Zhang, and Minlan Yu
Abstract: Programmable data plane has been moving towards deployments in data centers as mainstream vendors of switching ASICs enable programmability in their newly launched products, such as Broadcom’s Trident-4, Intel/Barefoot’s Tofino, and Cisco’s Silicon One. However, current data plane programs are written in low-level, chip-specific languages (e.g., P4 and NPL) and thus tightly coupled to the chip-specific architecture. As a result, it is arduous and error-prone to develop, maintain, and composite data plane programs in production networks. This paper presents Lyra, the first cross-platform, high-level language & compiler system that aids the programmers in programming data planes efficiently. Lyra offers a one-big-pipeline abstraction that allows programmers to use simple statements to express their intent, without laboriously taking care of the details in hardware; Lyra also proposes a set of synthesis and optimization techniques to automatically compile this “big-pipeline” program into multiple pieces of runnable chip-specific code that can be launched directly on the individual programmable switches of the target network. We built and evaluated Lyra. Lyra not only generates runnable real-world programs (in both P4 and NPL), but also uses up to 87.5% fewer hardware resources and up to 78% fewer lines of code than human-written programs.
Speaker Bio: Ennan Zhai is currently a Staff Engineer at Alibaba Group. His research focuses on building secure and reliable networking systems. Specifically, his work takes advantage of an interdisciplinary approach, integrating areas including verification, programming languages, database and security. Prior to joining Alibaba, he was a research scientist and lecturer at Yale University, where he also received his PhD in 2015.