Ken Calvert is Division Director for Computer and Network Systems in the Computer and Information Science and Engineering (CISE) Directorate at the National Science Foundation. He is there on rotation (since May 2016) from the University of Kentucky, where he is Gartner Group Professor in Network Engineering and former Chair of the Department of Computer Science. He has made contributions in various areas including network topology modeling, active/programmable networks, and future Internet architecture. He received his Ph.D. in computer science from the University of Texas at Austin. He holds a M.S. in computer science from Stanford University and a B.S. in computer science and engineering from the Massachusetts Institute of Technology. Prior to his appointment at the University of Kentucky, he was a Member of the Technical Staff at Bell Laboratories in Holmdel, NJ, and served on the faculty in the College of Computing at the Georgia Institute of Technology. He is an IEEE Fellow and a member of the ACM.
Talk Title: The Future of Computing and Network Systems - A View from NSF
Computing and communication infrastructure has changed dramatically over the past two decades, and there is no indication that this evolution will slow down. At the same time, the applications running on that infrastructure have become ever more critical to modern society. This talk will describe NSF activities related to advancing the state of the art, and will include some personal observations on challenges and opportunities.
Dirk Kutscher is a Professor for computer science and networking at the University of Applied Sciences Emden/Leer. Previously he has been the CTO for Virtual Networking and IP at Huawei's German Research Center. He is co-chairing two Research Groups in the Internet Research Task Force (IRTF) on Information-Centric Networking (ICNRG) and on Decentralized Internet Infrastructure (Proposed DINRG). Dirk has published several IETF RFCs, books, and research publications on Internet technologies. He has a PhD from Universität Bremen. Previously Dirk was the Chief Researcher for Networking at NEC Laboratories Europe and worked as a Visiting Researcher at KDDI R&D Laboratories in Japan.
Talk Title: Compute-First Networking (CFN): New Perspectives on Integrating Computing and Networking
Integrating computing with networking is considered promising if not essential for delivering adequate performance and efficiency for many exciting new use cases. For example, networked AR suggests that AR support functions could assist a user application by performing certain tasks (such as face recognition) that are too compute-intensive for some mobile devices. Low latency requirements demand that such functions be located close to the rendering devices, i.e., in the network, not at a far-away data center. As another example, predictive maintenance in smart manufacturing requires processing large amounts of data. Sometimes the volume may be too large to allow for real-time transmission to cloud-based analytics and processing functions. Instead, the data has to be stored and processed locally, i.e., close to generation. This processing might involve several layers of computations and intermediate results may be input to higher layer analytics.
Most technical approaches towards integrating such computing tasks into networks are based on the notion that computing and networking are two separate domains: for example, computation would take place in virtual machines, and the network (using IP forwarding, transport and application layer protocols) would simply serve as a connector of a set of virtual machines (in a virtual network or access network). This is based on the notion that instantiating and executing computation is a relatively heavy-weight and costly process. In recent years though, microservice architecture and the development of light-weight compute virtualization and isolation approaches suggest that there may be different ways to design in-network applications.
This talk introduces Compute-First Networking (CFN) -- an in-network computing approach that is based on the joint optimization of networking and computing resources. In CFN, applications are (at development time) split into components that can be embedded in networks at run-time in different ways, considering application properties, user preferences and operator policies. The embedding can dynamically change, according to dynamic network characteristics, nature and state of computations and data availability. For example, in a specific environment, it may be beneficial to not simply connect a user application to an already running compute instance behind a congested network link. Instead, the function could be moved and the connectivity graph could be updated accordingly. The talk will present an overview of the CFN architecture, design considerations and present new use cases for in-network computing that CFN enables.