wiki:GEC21Agenda/EveningDemoSession

Version 27 (modified by xuan.liu@mail.umkc.edu, 10 years ago) (diff)

--

  1. GEC 21 Evening Demo Session
    1. Location
    2. Schedule
    3. Session Leaders
    4. Details
    5. Directions and Logistics
    6. Projects
      1. Infrastructure and Measurement
        1. GENI Shakedown Expimeriments -- Intelligent Data Movement Service (IDMS)
        2. Networked Virtual Reality Based Training
        3. VTS VPN Support
      2. Experiments and Education
        1. GENI for the Masses: Demo of GENI MOOC Project
        2. Size-based Flow Management Enabling Dynamic DMZ
        3. PrimoGENI Constellation for Hybrid Network Experimentation
        4. Software Switch Data Plane Performance Characterization
        5. Network Complexity Index
        6. Hadoop Network Overhead Characterization
        7. Network Debugging
        8. Paradrop
        9. Live Transcoding Video on a Drone
        10. Hadoop-in-a-Hybrid-Cloud

GEC 21 Evening Demo Session

Location

Cyberinfrastructure Building
2709 East 10th Street
Bloomington, IN 47408
​http://it.iu.edu/cib/

Schedule

Tuesday Oct 21, 5:30p - 7:30p

Session Leaders

Heidi Picher Dempsey
GENI Project Office		 Manu Gosain
GENI Project Office		 Peter Stickney
GENI Project Office

Details

The evening demo session gives new and existing GENI experimenters and projects a chance to share their work in a live network environment. Demonstrations run for the entire length of the session, with teams on hand to answer questions and collaborate. This page lists scheduled demonstrations categorized in broad interest groups.

Directions and Logistics

Please visit Directions and Logistics for attendee and presenter logistics information.

Projects

Infrastructure and Measurement

GENI Shakedown Expimeriments -- Intelligent Data Movement Service (IDMS)

Intelligent data management service is dynamic storage and networking solution. It is deployed in GENI to evaluate performance. In this demonstration, we are extending our previous stitched shared-vlan experiment with meso-scale openFlow experiment spread across uh-eg, gpo-eg and fiu-eg. Two experiment are connected to each other using Layer 3 router deployed in VM at Illinois-ig. We will also be deploying Phoebus Gateways on backbone for external access to the experiment.

Participants:

Networked Virtual Reality Based Training

This demonstration focuses on a GENI based collaborative Virtual Reality based training environment for orthopedic surgery. A network based collaborative approach has been implemented outlined which enables distributed training of medical students from different locations. The virtual surgery environment is being remotely operated with the aid of a redundancy server.

Participants:

VTS VPN Support

The demo will show how to connect your own desktop/laptop directly into the dataplane of your slice using VTS VPN support. We will detail how to set up routing tables, L2 bridging, address MTU issues, etc. There will be demonstrations of this capability for multiple client platforms (at least windows and linux), and an example of a selectively-routed service where some of your traffic goes to your GENI slice and the rest goes through your existing internet connection.

Participants:

Experiments and Education

GENI for the Masses: Demo of GENI MOOC Project

We will demo an experiment-based Massive Open Online Course (MOOC) on the subject of computer networks, with lab experiments that run on GENI resources. This course is aimed at beginners who want to learn about how the Internet works, students who want an introduction to some research topics in networking, and instructors who may use these browser-based experiments as in-class demonstrations or homework assignments. The first course module, on computer routing, opens at the end of October; interested participants should stop by to see the demo and learn how to register. We will also show a poster on our WiMAX-based lab experiments, which may be of interest to instructors.

Participants:

Participants:

Size-based Flow Management Enabling Dynamic DMZ

Problem description: The current networking solutions for cybersecurity adopt static policies. Packets from and to supercomputers go through a DPI (deep packet inspection) device for security inspection before being routed toward their destinations. In a network configuration where the bandwidth of the links is equal to 10 gbps and the speed of the DPI is equal to 3 gbps, this device throughput represents the system bottleneck, bringing the bandwidth of the path down to 3 gbps. The adoption of an OpenFlow router allows us to adapt the cybersecurity policies, by dynamically managing the flow forwarding rules. We divide the flows generated by the supercomputers into two groups based on flow sizes: one group consists of elephant flows (the flow size is larger than a threshold) and the other group contains mice flows (the flow size is smaller than a threshold.) The goals of our dynamic flow management are to achieve higher network performance and to gain higher utilization of the network security mechanism. Demo: The client sends multiple flows to the server. The flows are forwarded by a software switch (OpenFlow switch) to the DPI first. Flows sent to the DPI are inspected for network security purposes. When the software switch detects a legitimate elephant flow, the elephant flow will be forwarded directly to the server bypassing the DPI. In this case, we avoid sending the elephant flow through the network bottleneck of the DPI to achieve higher network performance. By our dynamic flow management, we are able to monitor how the multiple flow entries are set up in the software switch flow table and how the flows are rerouted in the network. Equipment: software switch (OpenVSwitch on Ubuntu), Server (iperf on Windows 7), Client (iperf on Windows 7), DPI(OpenVSwitch on Ubuntu), OpenFlow controller(POX on Ubuntu).

Participants:

PrimoGENI Constellation for Hybrid Network Experimentation

PrimoGENI Constellation allows conducting hybrid, distributed, and at-scale network experiments on GENI resources, combining physical, simulated, and emulated networks. This demonstration will showcase new PrimoGENI capabilities for network researchers to conduct network experiments on various GENI resources, including InstaGENI and ExoGENI racks. This demonstration will also present the new prototype for a public model repository through which experimenters can create, reuse, extend, and share models for network experiments using PrimoGENI.

Participants:

Software Switch Data Plane Performance Characterization

The demo is on research with the data plane acceleration methods and their testing methods for software switches. Test planning with OpenFlow-capable software switches has various challenges. The control plane issues may hinder some of the test cases as they have been practiced for non-OpenFlow switches. We investigate solutions to such challenges and report on mapping of test plans on software switch performance characterization. Since most experimental setups utilize software switches for SDN experiments, the results will be applicable to all such SDN application research.

Participants:

Network Complexity Index

We investigate how network complexity index may help determine load balancing and performance guarantee issues in content delivery networks. A content request and response may yield a complexity understanding of the network that is very different from how network nodes are physically connected to each other. Such a representation may lead to better optimizations on the content reachability with load balancing guarantees. The experiment setup will help investigate feasibility of such insight into content delivery networks.

Participants:

Hadoop Network Overhead Characterization

The demo is on experimental investigations of overhead in hadoop's network usage while ensuring resiliency in the application layer. We investigate the coordination information exchange in hadoop's typical processing. We report on how such overhead changes with process characteristics and how failures may increase such exchange traffic. A measurement setup on GENI is used to conduct this research.

Participants:

Network Debugging

The demo is on adaptive classification of incoming traffic using OpenFlow protocol on an SDN overlay into a wireshark monitoring site. The network engineers may enter filtering instructions to further introduce zooming into specific traffic and identify source and destination. The demo is on the feasibility and completeness of such a network debugging setup in an enterprise network. A presumed hierarchy of access, edge and aggregation nodes has been emulated on GENI testbed to realize this experiment.

Participants:

Paradrop

We will demo the Paradrop Platform, which is a software platform that allows developers to launch applications onto specialized Access Points that exist in the home. This provides the ability to introduce unique control and high quality value adds onto services the end-user chooses to use in their home including applications related to Internet of Things, high-definition media content distribution, and others. For this demo, we will showcase the Platform's ability to dynamically launch and control virtual machines that are running within the Access Point for a few specific services.

Participants:

Live Transcoding Video on a Drone

The WiNGS lab is working to set up a city wide testbed utilizing WiMAX base stations for connectivity to many client devices. This demo shows our drone we use to transcode live video we send over the WiMAX connection. We are looking at using this technology to assist emergency response vehicles with real time updates while they are on route.

Participants:

Hadoop-in-a-Hybrid-Cloud

MapReduce is a programming model for processing and generating large data sets, and Hadoop, a MapReduce implementation, is a good tool to handle Big Data. Cloud computing with its ubiquitous characteristic, on demand and dynamic resource provisioning at low cost has potential to be the environment to treat big data. However, using Hadoop on the cloud spends time and requires technical knowledge from users. The hybrid cloud leverages these requirements, because it’s necessary to evaluate the resources in private cloud and, if necessary, obtain and prepare on-demand resources in the public cloud. Moreover, the simultaneous management of private and public domains requires an appropriate model that combines performance with minimal cost. We propose an architecture to make the orchestration of Hadoop applications in hybrid clouds, which include a private cloud at Unicamp and UMKC, and GENI as the public cloud. The core of the model consists of a web portal for submissions, an orchestration engine and an execution services factory. Through these three components it’s possible to automate the preparation of a cross-domain cluster, performing the provisioning of files involved, managing the execution of the application, and making the results available to the user. In this demo, we will show the web portal interface and how to use this portal to create VMs and initialze it as a Hadoop worker.

Participants:

Attachments (19)