Seminars & Colloquia
Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University
"Large-scale network prototyping based on lightweight virtualization techniques"
Monday June 30, 2014 11:00 AM
Location: 3300, EBII NCSU Centennial Campus
(Visitor parking instructions)
Network simulation is an effective method for network operation analysis tasks, such as estimation of different indices of network performance (throughput, end-to-end delay, etc.), verify that network resources are well balanced for different workloads, understanding what malicious code can do to a specific network, the impact of changes in the network topology and resources allocation, and many others. These are especially important in Large Scale Networks (LSN), a term we use to indicate networks with no fewer than 10,000 nodes. Traditionally, such networks have been modeled with queuing theory techniques using statistical averaging of many network operation parameters. Averaging was used to simplify the model by reducing its size. The price of this simplification is a precision in the prediction of real network operation. Our approach to SDN network emulation based on light virtualization techniques is called the Hi-Fi approach. The Hi-Fi approach was proposed to the network environment emulation for the SDN controller application development. The proposed approach allows an SDN network emulator to be developed with high level of confidence.
In this talk, we propose a way the Hi-Fi approach can be used for LSN simulation for both SDN and traditional networks such as TCP/IP based ones. Of the several components of a practical simulation system, we focus on the following important ones: run-time environment organization and modeling time management subsystem. During the presentation, I will talk about the scalable Network Prototyping Simulator (NPS), which allows us to surmount the problems of LSN simulation. NPS is based on OS Linux Lightweight Virtualization Containers (LXC) techniques. The approach used in NPS allows achieving high fidelity simulation results, close to network emulationl but unlike the emulation approach, NPS scales to LSN networks.
Antonenko Vitaly is a Ph. D. student in the Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University, currently on a research visit hosted by the Renaissance Computing Institute, UNC Chapel Hill. From 2005 to 2010 he studied as a student at the Moscow State University. From October 2010 until present time he has been a PhD student and researcher in Computer Systems Laboratory (LVK lab), Moscow State University. His main research interests are in the areas of network modeling, computer viruses, net-worms, DDoS-attacks, network protocols, IDS, IPS, firewalls, antiviruses, network security, malware propagation, simulation modeling. Antonenko has taken part in several scientific projects in LVK lab in the areas of simulation modeling and information security. He also participates in the University studying process, as a lecturer of â€œpractical work in computer networksâ€? seminar for students. He is currently an active contributor to the GENI national testbed effort funded by the US National Science Foundation.
Host: Rudra Dutta, CSC