Opened 8 years ago

Closed 8 years ago

#860 closed (fixed)

GEC13 Poster Request: Adaptive Source Routing on GENI

Reported by: Oliver Michel Owned by: agosain@bbn.com
Priority: major Milestone:
Component: GPO Version: SPIRAL4
Keywords: GEC13 Cc:
Dependencies:

Description

We would like to present one normal-size poster at the GEC13 about the Adaptive Source Routing of the University of Illinois at Urbana-Champaign. The primary investigator is Professor Brighten Godfrey. This is the project's description:

Today's routing techniques on the Internet rely completely on decisions within the network. Lacking an end-to-end view, routing algorithms therefore often react slowly to dynamics in the network, and do not take into account the type of traffic being routed. For example, traffic with the demand of high bandwidth is not distinguished from services needing low latency such as real-time applications.

This project is investigating the alternate approach of source-controlled routing (SCR). Assuming a source node holds information about possible routes to destination nodes including information about the characteristics of the different paths obtained by continuous probing, a source can decide which path along the network a particular packet should traverse for optimal results. In addition to that, multipath routing can be used to dramatically decrease latency and routing-related delay with a reasonable overhead on traffic when sending packets simultaneously via different links, using only the packet which arrives first at the destination.

Previously, we have presented at GENI results of measurements of the effectiveness of SCR in a PlanetLab overlay network, as well as a simple demo of the technique on GENI's OpenFlow network. This poster will present measurements of SCR in GENI's OpenFlow network, and will compare them with our conclusions on PlanetLab. In particular we are measuring the effectiveness of various path selection algorithms and multipath strategies, leveraging the GENI OpenFlow network to decrease unpredictability in the previously used overlay deployment. Our results from the overlay deployment and first results from the GENI topology show that the latency-distribution's tail can be significantly decreased when using source-controlled routing with different path-selection strategies.

Change History (5)

comment:1 Changed 8 years ago by Oliver Michel

Summary: Adaptive Source Routing on GENIGEC13 Poster Request: Adaptive Source Routing on GENI

We would like to present one normal-size poster at the GEC13 about the Adaptive Source Routing Project at the University of Illinois at Urbana-Champaign. The primary investigator is Professor Brighten Godfrey. This is the project's description:

Today's routing techniques on the Internet rely completely on decisions within the network. Lacking an end-to-end view, routing algorithms therefore often react slowly to dynamics in the network, and do not take into account the type of traffic being routed. For example, traffic with the demand of high bandwidth is not distinguished from services needing low latency such as real-time applications.

This project is investigating the alternate approach of source-controlled routing (SCR). Assuming a source node holds information about possible routes to destination nodes including information about the characteristics of the different paths obtained by continuous probing, a source can decide which path along the network a particular packet should traverse for optimal results. In addition to that, multipath routing can be used to dramatically decrease latency and routing-related delay with a reasonable overhead on traffic when sending packets simultaneously via different links, using only the packet which arrives first at the destination.

Previously, we have presented at GENI results of measurements of the effectiveness of SCR in a PlanetLab overlay network, as well as a simple demo of the technique on GENI's OpenFlow network. This poster will present measurements of SCR in GENI's OpenFlow network, and will compare them with our conclusions on PlanetLab. In particular we are measuring the effectiveness of various path selection algorithms and multipath strategies, leveraging the GENI OpenFlow network to decrease unpredictability in the previously used overlay deployment. Our results from the overlay deployment and first results from the GENI topology show that the latency-distribution's tail can be significantly decreased when using source-controlled routing with different path-selection strategies.

comment:2 Changed 8 years ago by agosain@bbn.com

Status: newassigned

Thank you for your request.

comment:3 Changed 8 years ago by agosain@bbn.com

(If you need to respond to something in this message, please update the ticket on the wiki, or send mail to agosain@bbn.com and I can update it for you. Replies to trac@geni.net will return an auto-response.)

We can now confirm that the resources you requested will be available for your demo on Wednesday evening.

comment:4 Changed 8 years ago by agosain@bbn.com

ERRATA

We can now confirm that the resources you requested will be available for your demo on 3/13 Tuesday evening.

comment:5 Changed 8 years ago by agosain@bbn.com

Resolution: fixed
Status: assignedclosed

Thanks again for participating in the GEC 13 evening demo session.

We've added information about the demo session to http://groups.geni.net/geni/wiki/GEC13Agenda/EveningDemoSession. Please feel free to edit the section about your demo, add any participants who you'd like to credit, add links to other relevant pages, and attach any materials (slides, PDFs of posters, etc) about your demo.

If you have any questions or concerns, or need a hand with any of that, just drop a note to help@geni.net.

Note: See TracTickets for help on using tickets.