|09:15||Demand-aware and distributed cloud networking: Let’s get physical!
Stefan Schmid (University of Vienna, Austria)
|10:00||Morning coffee break|
|10:30||Machine Learning Training: Research Challenges and Opportunities for Distributed Computing
Giovanni Neglia (Inria Sophia Antipolis, France)
|11:15||Towards autonomous industries
Robert Birke (ABB Corporate Research, Switzerland)
|14:00||Resilience at the Edge
Ivona Brandić (Vienna University of Technology, Austria)
|14:45||End-to-end Orchestration Automation in Distributed Cloud: Resource, Service and Multi-stakeholders Aspects
Róbert Szabó (Ericsson Research, Hungary)
|15:30||Practical Network Monitoring in Software Switches
Roy Friedman (Technion – Israel Institute of Technology, Israel)
|16:15||Afternoon coffee break|
|16:45||The Distributed Cloud: Theory Meets Practice
|Robert Birke received his PhD degree from the Politecnico di Torino in 2009 with the telecommunications group under the supervision of professor Fabio Neri. Currently he works as a senior research scientist at ABB Corporate Research, Switzerland. Before that, he was a postdoc in the cloud server technologies group at IBM Zurich Research Lab, Switzerland. He is a co-author of more than 80 scientific papers. His main research interests are applied machine learning, cloud computing, and networking with special focus on performance, quality of service, and virtualization. Dr. Birke is a senior IEEE member.|
|Talk title: Towards autonomous industries|
|Talk abstract: Digitalization and autonomy are shaping the fourth industrial revolution. This revolution relies on many technological corner stones including computing, e.g., fog computing, networking, e.g., 5G, and algorithms, e.g., deep learning. Some of these technologies are shared with consumer applications, however industrial requirements and boundary conditions can significantly differ in many respects. This talk will present the requirements from an industrial perspective leading to more autonomous industries and pinpoint some of the main research challenges relevant to the industrial setting.|
|Ivona Brandic is University Professor for High Performance Computing Systems at the Institute of Information Systems Engineering, Vienna University of Technology (TU Wien). In 2015 she was awarded FWF START prize, the highest Austrian award for early career researchers. Since 2016 she is member of the Young Academy of the Austrian Academy of Sciences.
From 2002 to 2007 she was Assistant Professor at the Department of Scientific Computing, University of Vienna. Form 2007 to 2014 she was Assistant Professor at the Institute of Information Systems, TU Vienna. From 2014 to 2015 she was Assistant Professor at the Institute for Software Technology and Interactive Systems. She received her PhD degree in 2007 and her venia docendi for practical computer science in 2013, both from Vienna University of Technology. From June to August 2008 she was visiting researcher at the University of Melbourne, Australia. I. Brandic is on the Editorial Board of the IEEE Transactions on Cloud Computing, IEEE Magazine on Cloud Computing and IEEE Transactions on Parallel and Distributed Systems. In 2011 she received the Distinguished Young Scientist Award from the Vienna University of Technology for her project on the Holistic Energy Efficient Hybrid Clouds. Her interests comprise virtualized HPC systems, energy efficient ultra-scale distributed systems, massive-scale data analytics, Cloud & workflow Quality of Service (QoS), and service-oriented distributed systems. She published more than 50 scientific journal, magazine and conference publications and she co-authored a text-book on federated and self-manageable Cloud infrastructures. She has been serving more than 50 program committees among others Supercomputing, CCGrid, CloudCom, EuroPar, and COMPSAC.
|Talk title: Resilience at the Edge|
|Talk abstract: Internet of Things (IoT) is revolutionizing the way how information is processed and stored. Due to latency sensitive applications and huge amounts of data produced at the edge of the network, more and more data is processed where it is produced – namely on the edge. This development results in completely new network topologies where besides massive data centers we experience growing amount of so called micro data centers on the edge of the network. However, increasing complexity of multiple data centers necessary to execute an application represents a new challenge for the deployment and runtime operation of large scale applications like those in the area of smart cities, self-driving vehicles and tele medicine. The challenge thereby is to deploy application in a way to satisfy user requirements in form of different Quality of Service parameters (e.g., latency) but at the same time minimize energy consumption necessary to execute the application. In this talk we discuss challenges considering resilience that arise when deploying near real time analytics on the edge of the network.|
|Roy Friedman is a professor in the department of Computer Science at the Technion – Israel Institute of Technology. His research interests include Network Streaming Protocols, Caching, Replication, Fault-Tolerance, Dependability, High Availability, Consistency, and Mobile Computing. Roy Friedman serves as an associate editor for the IEEE TDSC and PC co-chair or OPODIS 2019. In the past he served as PC co-chair for ACM DEBS 2015, ACM SYSTOR 2014 and Autonomics 2009 as well as vice chair for IEEE ICDCS 2013+2006 and EuroPar 2008+2003, and fast abstract chair for IEEE DSN 2013. He has published more than 150 papers and holds 3 USA patents. Formerly, Roy Friedman was an academic specialist at INRIA (France) and a researcher at Cornell University (USA). He is a founder of PolyServe Inc. (acquired by HP) and holds a Ph.D. and a B.Sc. from the Technion|
|Talk title: Practical Network Monitoring in Software Switches|
|Talk abstract: Recent trends such as Network Function Virtualization (NFV), Software Defined Networks (SDN) and multi-tenant virtual machine hosting imply that many network functions, which were traditionally realized in hardware, must now be implemented in software. Measuring and monitoring network statistics are extremely important functions, as they enable effective traffic engineering, load balancing, caching, and security. Yet, developing such capabilities in software is tricky, especially given the ever-increasing line rates of modern networks. While SRAM memory, which is fast enough to cope line rates, is larger in modern servers’ CPUs than in switches, it is still far from being abundant. On the other hand, software lacks the fine grain massive parallelism that hardware designs benefit from. This means that naïve implementations of sketches and similar traditional techniques do not perform well in software. In this talk I will exemplify the challenges in designing fast network measurements and monitoring functions in software. I will then present some recent results about how to effectively combine sampling and sketching techniques in order to meet these challenges.|
|Giovanni Neglia received the master’s degree in electronic engineering and the PhD degree in telecommunications from the University of Palermo, Italy, in 2001 and 2005, respectively. He has been a researcher at Inria, Sophia Antipolis, France, since September 2008. In 2005, he was a research scholar with the University of Massachusetts, Amherst, visiting the Computer Networks Research Group. Before joining Inria, he was a post-doctorate with the University of Palermo and an external scientific advisor with the Maestro Team at Inria. His research focus on modeling and performance evaluation of networks. He is area editor for the Elsevier Computer Communication journal (COMCOM), and associate editor for IEEE Trans. on Mobile Computing.|
|Talk title: Machine Learning Training: Research Challenges and Opportunities for Distributed Computing|
We explain why existing general-purpose computation frameworks like Spark are rarely used to train large machine learning models, and introduce parameter server and ring-allreduce, the most common paradigms for distributed machine learning training. We motivate the need for research from the distributed systems’ community and present some recent directions pursued in our team on adaptive backup workers and communication topology design.
|Stefan Schmid is a Professor at the Faculty of Computer Science, at University of Vienna, Austria. He obtained his diploma (MSc) in Computer Science at ETH Zurich in Switzerland (minor: micro/macro economics, internship: CERN) and did his PhD in the Distributed Computing Group led by Prof. Roger Wattenhofer, also at ETH Zurich. As a postdoc, he worked with Prof. Christian Scheideler at the Chair for Efficient Algorithms at the Technical University of Munich and at the Chair for Theory of Distributed Systems at the University of Paderborn, in Germany. From 2009 to 2015, Stefan Schmid was a senior research scientist at the Telekom Innovation Laboratories (T-Labs) and at TU Berlin in Germany (Internet Network Architectures group headed by Prof. Anja Feldmann). In 2013/14, he was an INP Visiting Professor at CNRS (LAAS), Toulouse, France, and in 2014, a Visiting Professor at UniversitÃ© catholique de Louvain (UCL), Louvain-la-Neuve, Belgium. From 2015 to 2017, Stefan Schmid was a (tenured) Associate Professor in the Distributed, Embedded and Intelligent Systems group at Aalborg University, Denmark, and continued working part-time at TU Berlin, Germany. Since 2015, he serves as the Editor of the Distributed Computing Column of the Bulletin of the European Association of Theoretical Computer Science (BEATCS), since 2016 as Associate Editor of IEEE Transactions on Network and Service Management (TNSM), and since 2019 as Editor of IEEE/ACM Transactions on Networking (ToN). Stefan Schmid received the IEEE Communications Society ITC Early Career Award 2016. His research interests revolve around the fundamental and algorithmic problems of networked and distributed systems.|
|Talk title: Demand-aware and distributed cloud networking: Let’s get physical!|
|Talk abstract: Networks from, to and inside datacenters have become a critical infrastructure of our digital society. In order to improve the efficiency of these networks, over the last years, researchers have put major efforts into increasing the flexibility of the underlying communication technologies, enabling demand-awareness on different layers, from the application layer down to the networking layer. Enabled by novel reconfigurable optical technologies, the physical layer now emerges as a next frontier where demand-awareness can be introduced.
In this talk, I will present the vision, enablers, motivation, metrics, and algorithmic challenges of demand-aware networks. In particular, I will present a distributed self-adjusting cloud network which provides route lengths that meet entropy lower bounds.
|Róbert Szabó, PhD, is a master researcher at Ericsson Research, Hungary since 2013. At Ericsson, he was the technical coordinator of the EU-FP7 integrated project: Unifying Cloud and Carrier Networks (UNIFY) /2013-2016/ and he was the project coordinator of the H2020 5G-PPP 5G Exchange (5GEx) innovation action /2015-2018/. Dr. Szabo has an associate professor position (part-time) at the Dept. of Telecommunications and Media Informatics (TMIT), Budapest University of Technology and Economics (BME). He was the president of the Telecommunications Section of the Scientific Association for Infocommunications (HTE), Hungary /2005-2007/. He was the deputy head of the TMIT, BME (2008-2010). He was the head of the High-Speed Networking Lab (HSNLab), at BME /2007-2012/. His researches were supported by the János Bolyai Scholarship of the Hungarian Academy of Science (MTA). He was a member of the Future Internet Award jury /2010- 2013/. He is the co-author of over 80 publications. He supervised the work of 3 graduated PhD students at TMIT, BME.|
|Talk title: End-to-end orchestration automation in distributed cloud: resource, service and multi-stakeholders aspects|
|Talk abstract: Edge computing provides compute and storage resources with adequate connectivity (networking) close to the devices generating / terminating traffic. The benefit is the ability to provide new services with strict requirements on, e.g., latency, bandwidth or on local break-out possibilities. Many use-cases for 5G (IoT, connected cars, Industry 4.0, …) span the device, access-, distributed-, national- or global sites. This requires a solution that can handle any workload, anywhere in the network, with end-to-end orchestration. Distributed cloud goes along with automated deployment of applications at just the right location in the network to optimize resource efficiency and user experience. Distributed clouds, however, may well span across multiple providers, both infrastructure and online service providers, and may include enterprises or end customer resources too. In such scenarios, autonomous systems shall self-organize themselves into situational orchestration structures (hierarchies) to enable end-to-end automation. Last but not least applications are built on top of multiple-layers of services, which may be managed on their own (XaaS paradigm). How distributed resources and managed services from multiple-stakeholders may be brought together for end-to-end service automation is discussed based on results and insights gained from a proof of concept research prototype.|