ACM 3rd International Workshop on System and Network Telemetry and Analytics (SNTA'20)

CFP: https://sdm.lbl.gov/snta/2020/SNTA2020-CFP.html
Program: https://sdm.lbl.gov/snta/2020
SNTA archive
Email: SNTA.help@gmail.com  

In conjunction with HPDC 2020, June 23-26, 2020, Virtual Conference via Zoom

Workshop Program

2pm, June 24, 2020, GMT
7am, June 24, 2020, PDT (USA)
9am, June 24, 2020, CDT (USA)
10am, June 24, 2020, EDT (USA)
3pm, June 24, 2020 (England)
4pm, June 24, 2020 (Italy, France, Sweden)
10pm, June 24, 2020 (Taiwan)
11pm, June 24, 2020 (Korea)
Zoom downloads
Zoom: https://lbnl.zoom.us/j/93555767613
Meeting ID: 935 5576 7613
* Zoom session is password protected.
* All SNTA/HPDC registered participants will receive the zoom access information.
* This zoom session is only for SNTA workshop 2020.
* HPDC has a separate zoom session info.

Thanks for your participation! Click here for the recorded SNTA2020 welcome session.
All video presentations will be available on YouTube SNTA 2020 channel.

Registration: Please register using HPDC registration.
Zoom connection details will be sent to the list of registered attendees.

The Q&A session for papers is done through Slack (https://snta-workshop.slack.com) channel (#snta20).
This channel will remain open, and you can join the SNTA slack channel through this invitation link:
https://join.slack.com/t/snta-workshop/shared_invite/zt-exna3bhz-HW5LpLjvlLDz1NLJyPlYbw.

Program order subject to change.
SNTA 2020 Co-Chairs: Massimo Cafaro, Jinon Kim, Alex Sim

Session 1

Keynote Presentation
    Inder Monga, Energy Sciences Network
    Title: Analytics-Driven Networking: When the Computer becomes the Network
(Slides), (Summary presentation), (Full presentation)
    Abstract: As the era of 'human-managed networking' passes to 'analytics-driven networking', more and more data about networks, including the constituent flows, is being tracked and retrieved. With networks now needing to be an effective sensor, new methods are being proposed to create and manage this streaming telemetry. While collection of this telemetry is happening at an unprecedented scale, it is unclear if the data is of enough resolution to make real-time decisions needed for fine-grained control, or for the application of new machine learning/artificial intelligence techniques. New techniques are being developed to provide high-precision telemetry as new analytics to take advantage of that are being developed.
While telemetry is being used to provide insights about networks, applications want more instrumentation from the currently opaque network. Currently, there is little ability for the applications to interact with the network to exchange information, negotiate performance parameters, discover desired performance metrics, or receive status/troubleshooting information in real time.
This talk will share challenges, opportunities and new research in progress, from a science network view, as we evolve the network to an analytics-driven, autonomous system. In addition, the talk will discuss the new nation-scale testbeds being built, with new capabilities, like in-network processing, that could revolutionize how we measure and process end-to-end workflows.

Indermohan (Inder) S. Monga serves as the Division Director for Scientific Networking Division, Lawrence Berkeley National Lab and Executive Director of Energy Sciences Network, a high- performance network user facility optimized for large-scale science, interconnecting the National Laboratory System in the United States. Under his leadership, the organization focuses on advancing the science of networking for collaborative and distributed research applications. He contributes to ongoing research projects tackling network programmability, analytics and quality of experience driving convergence between application layer and the network. He currently holds 23 patents and has 20+ years of industry and research experience in telecommunications and data networking. His work experience in the private sector includes network engineering for Wellfleet Communications and Nortel where he focused on application and network convergence. His undergraduate degree is in electrical/electronics engineering from Indian Institute of Technology in Kanpur, India, with graduate studies from Boston University.     

Session 2
















(1112)

KDetect: Unsupervised Anomaly Detection for Cloud Systems Based on Time Series Clustering
(Slides), (Summary presentation), (Full presentation)
    Swati Sharma, Amadou Diarra (University of Grenoble Alpes, Grenoble), Frederico Alvares (Easyvirt), Thomas Ropars (University of Grenoble Alpes, Grenoble)

Data-driven Learning to Predict WAN Network Traffic
(Slides), (Summary presentation), (Full presentation)
    Nandini Krishnaswamy, Mariam Kiran (LBNL), Kunal Singh (UC Berkeley), Bashir Mohammed (LBNL)

Feature Selection Improves Tree-based Classification for Wireless Intrusion Detection
(Slides), (Summary presentation), (Full presentation)
    Shilpa Bhandari, Avinash Kukreja, Alina Lazar (Youngstown State University), Alex Sim, Kesheng Wu (LBNL)

Using Machine Learning for Intent-based Provisioning in High-Speed Science Networks
(Slides), (Summary presentation), (Full presentation)
    Hocine Mahtout (Bordeaux Graduate School of Engineering), Mariam Kiran (LBNL), Anu Mercian (HP Labs), Bashir Mohammed (LBNL)

Session 3

Keynote Presentation II
    Eric Chan-Tin, Loyola University, Chicago
    Title: 2020 Vision for Web Privacy
(Slides), (Summary presentation), (Full presentation)
    Abstract: Privacy is getting eroded as more surveillance is happening. The history of web privacy will be discussed along with a vision for the future. This talk will discuss how web users are tracked, what can be done about it, the impact of web surveillance on society, and privacy regulations.
Web privacy has evolved significantly in the past two decades. There was no application-level encryption and snooping on what users were doing was trivial. As encryption, proxies, and other anonymity schemes start getting used, web surveillance has switched to using metadata. At the moment, achieving web privacy is similar to the arms race of virus/anti-virus. New methods to break web users' privacy are found as current vectors are patched or mitigated.
Web surveillance or invading web users' privacy can have both useful and harmful consequences. The methods can be used to identify illegal activity and ensure that users are who they say they are. However, these methods can also be used to perform censorship and track what users are doing on the web without their knowledge. The latter could be used for targeted advertising - on one hand, this means more useful advertisements, but on the other, it could mean targeted pricing.
Many people might say they have nothing to hide; this talk will show how this is such a misleading myth. The evolution and future of webbrowser fingerprinting and website fingerprinting will be discussed, along with other societal and human impacts of web privacy on the future.

Eric Chan-Tin is an assistant professor in the Department of Computer Science and POC for the Center of Cybersecurity at Loyola University Chicago. Previously, he was an associate professor in the CS Department at Oklahoma State University. He received his Ph.D. degree from the University of Minnesota in 2011 and his B.A. from Macalester College in 2006. His research areas are in network security, distributed systems, privacy, and anonymity. He has published over 30 peer-reviewed papers, including publications at conferences and journals such as ACM CCS, NDSS, ACM TISSEC, and IEEE TIFS.     

Session 4

















(1122)

HPC Workload Characterization Using Feature Selection and Clustering
(Slides), (Summary presentation), (Full presentation)
    Jiwoo Bang, Chungyong Kim (SNU), Kesheng Wu, Alex Sim, Suren Byna (LBNL), Sunggon Kim, Hyeonsang Eom (SNU)

Transfer Learning Approach for Botnet Detection based on Recurrent Variational Autoencoder
(Slides), (Summary presentation), (Full presentation)
    Jeeyung Kim, Alex Sim (LBNL), Jinoh Kim (Texas A&M University-Commerce), Kesheng Wu (LBNL), Jaegyoon Hahm (KISTI)

Finding the Optimal Reconfiguration for Network Function Virtualization Orchestration with Time-varied Workload
(Slides), (Summary presentation), (Full presentation)
    Satyajit Padhy, Jerry Chou (National Tsing Hua University)

Evaluation of Deep Learning Models for Network Performance Prediction for Scientific Facilities
(Slides), (Summary presentation), (Full presentation)
    Makiya Nakashima (Texas A&M University-Commerce), Alex Sim (LBNL), Jinoh Kim (Texas A&M University-Commerce)