FiCloud

Keynote and Industry Talks

Keynote 1: High-Performance Data Center Networking

Prof. Minlan Yu

Harvard University, USA

Abstract: As data center networks strive to provide high throughput and ultra-low latency, they are increasingly sensitive to many fine timescale events such as microbursts, packet losses, and high queuing delay. It is challenging to capture these events because it requires microsecond-level counters to capture transient network states and high overhead to capture many such events in large networks. Today, without information about these fine timescale events, we have to infer network states and design complex heuristics for control decisions (e.g., congestion control). Moreover, due to the lack of precise information about these events, applications often suffer from tail latency problems caused by these events and struggle to locate the root causes. To address these challenges, we build network telemetry systems that can capture flow-level and packet-level events at fine timescale at both hosts and switches with low overhead. Such a telemetry system then provides a data foundation for us to design precise control solutions that quickly react to fine timescale events and diagnosis systems that can enable debugging large-scale applications with detailed information and low overhead. In this talk, we will discuss a few measure and control systems we built in my group to illustrate the design. Some of our work has been deployed in production data centers and adopted by switch/NIC vendors.

Biography: Minlan Yu is a Gordon Mckay professor at Harvard School of Engineering and Applied Science. She received her B.A. in computer science and mathematics from Peking University in 2006 and her M.A. and PhD in computer science from Princeton University in 2008 and 2011. Her research interests include data networking, distributed systems, enterprise and data center networks, and software-defined networking. She received the ACM SIGCOMM doctoral dissertation award in 2011 and NSF CAREER award in 2015. She served as PC co-chair for NSDI, HotNets, and several other conferences and workshops.

Keynote 2: Cloud Intelligence/AIOps – Infusing AI/ML into Large-scale Cloud Systems

Qingwei Lin

Microsoft Research, Asia

Abstract: In the past fifteen years, the most significant paradigm shift in the computing industry is the migration to cloud computing, which brings unprecedented opportunities of digital transformation to business, society, and human life. Therefore, the quality of cloud platforms, including reliability, performance, efficiency, security, sustainability, etc., has become immensely important. However, the distributed nature, massive scale, and high complexity of cloud platforms present huge challenges to design, build, and operate such systems effectively and efficiently. To address these challenges, "Cloud Intelligence/AIOps" is to infuse AI/ML into the designing, building, and operation of high-quality and high-efficiency cloud systems at scale. In this talk, I will first introduce the concept of “Cloud Intelligence/AIOps” and its research landscape. Then using a few projects at Microsoft as examples, I will talk about the work from Microsoft Research and its impact. I will also discuss the research challenges and opportunities in Cloud Intelligence/AIOps moving forward.

Biography: Qingwei Lin is a Sr. Principal Research Manager at the DKI (Data, Knowledge, Intelligence) area of Microsoft Research Asia. He is leading a team of researchers working on machine learning and data mining technologies for Cloud Intelligence/AIOps. In Cloud Intelligence/AIOps area, Qingwei has ~50 publications in influential conferences such as OSDI, NSDI, ICSE, FSE, AAAI, IJCAI, SigKDD, etc. The research technologies have been transferred into multiple Microsoft products, such as Azure, Office, Windows, etc. Qingwei chaired Microsoft company-wide "Cloud Service Intelligence Summit" for 4 consecutive years. He joined Microsoft Research in 2006.

Keynote 3: Leveraging Cloud, Fog and Mist Computing for Real-Time Applications: A Resource Allocation and Scheduling Perspective

Prof. Helen D. Karatza

Aristotle University of Thessaloniki, Greece

Abstract: The ongoing expansion of the Internet of Things (IoT) has led to the emergence of new computing paradigms, such as fog and mist computing, in order to address the inherent latency of the remote cloud resources. The vast amount of data generated by IoT sensors and devices typically requires processing in a real-time manner, which cloud resources cannot usually provide due to their physical distance from the IoT layer. Fog computing extends the cloud closer to where the IoT data are generated in an attempt to minimize latency. Mist computing, a lightweight form of fog computing, extends the fog layer even closer to the IoT sensors and devices. The collaboration of mist, fog and cloud resources for the processing of real-time applications involves many challenges. Particularly important is the effective resource allocation and scheduling of the real-time workload on the multi-tier resources. In this talk, we will shed light on resource allocation and scheduling techniques for real-time applications, leveraging the power of cloud, fog and mist computing. Recent trends and novel approaches will be presented. In the conclusion, we will explore future research directions.

Biography: Helen Karatza (Senior member, IEEE, ACM, SCS) is a Professor Emeritus in the Department of Informatics at the Aristotle University of Thessaloniki, Greece. Her research interests include cloud and fog computing, resource allocation and scheduling, real-time distributed systems, simulation and performance evaluation of large-scale distributed systems. She has authored or co-authored more than 200 technical papers and book chapters including five papers that earned best paper awards at international conferences. She served as an elected member of the Board of Directors at Large of the Society for Modeling and Simulation International. She served as Chair and Keynote Speaker in international conferences. She is the Editor-in-Chief of the Elsevier journal “Simulation Modelling Practice and Theory” and member of the Editorial Board of the “Future Generation Computer Systems” Elsevier journal. She was Editor-in-Chief of “Simulation Transactions of the Society for Modeling and Simulation International”, Associate Editor of “ACM Transactions on Modeling and Computer Simulation” and Senior Associate Editor of the “Journal of Systems and Software” of Elsevier. She served as Guest Editor in numerous Special Issues of international journals.

Keynote 4: Reinforcement Learning for Service Placement and Resource Provisioning in Mobile Edge Computing

Prof. Jamal Bentahar

Concordia University, Canada

Abstract: In the recent context of 6G and the Internet of Everything (IoE), more computing resources are required. Mobile Edge Computing (MEC) provides an efficient framework to deal with this problem. This talk will present an intelligent and proactive resource provisioning and service placement solution that considers the dynamic changes of service demands, the limitation of available computing resources of MEC, and the increase in the number and complexity of IoE services. The solution introduces a deep reinforcement learning algorithm where multiple requirements are considered such as the prediction of the resource usage of scaled applications, the prediction of available resources by hosting servers, as well as making service placement decisions. The solution addresses the long learning time for the algorithm to converge. The talk will also present a reinforcement leering solution to the problem of minimizing both, the network delay, which is the main objective of MEC, and the number of edge servers to provide a MEC design with minimum cost. This MEC design consists of edge servers placement and base stations allocation, which makes it a joint combinatorial optimization problem. Experiments and simulation results will be discussed. .

Biography: Jamal Bentahar is a Professor with Concordia Institute for Information Systems Engineering at Concordia University, Canada. He received the Ph.D. degree in computer science and software engineering from Laval University, Canada, in 2005. He obtained in 2006 the highly competitive NSERC Postdoctoral Fellow at Simon Fraser University, Canada. His research interests include artificial intelligence, machine learning, reinforcement learning, multi-agent systems, cloud/edge computing, computational logics, model checking, and applied game theory. He served as co-chair of the NSERC evaluation group from 2016 to 2018. He has published more than 200 papers in competitive venues such as AAMAS, IJCAI, AAAI, ICSOC, SCC, ICWS, IEEE TSC, ACM TIST, FGCS.