This is a "tour" of one of the demos done at the DARPA FTN Program Meeting held in January 2001 at St. Petersburg, Florida. In this demo, we demonstrated that we have implemented pricing at the level of RSVP, COPS, and the policy server / policy database. Users requesting bandwidth with RSVP will be admitted or denied partly on their willingness to pay the "going price" of the bandwidth. In this demo, we created a web interface that allows controlling the network, initiating requests and applications, and viewing the results.
The test network had 5 nodes. Two nodes were connected by a "bottleneck" link with 10Mbps of bandwidth. The Realplayer video streaming application was executed across this bottleneck link, between a client and a Realplayer server. The application continually played a 2 minute video of highlights from "The Phantom Menace". The two ways in which performance can be evaluated are a) by viewing the video and subjectively judging its quality, and b) by looking at a graph of the throughput rate, which is provided by the "statistics" function built in to Realplayer. On this web page we show the "screenshots" of the demo execution, captured to disk.
In the first demo, we started the realplayer video
with no other traffic on the bottleneck link. The video playback was smooth,
and the bandwidth provided was almost 100% of the requested amount. The screenshots
below show the web interface, the video, and the statistics window.
In the second demo, 10 Mbps of background traffic was transmitted over the bottleneck link, in addition to the video. No bandwidth was reserved for the video. The video rate was reduced to 12 Kbps from 152 Kbps, and the throughput was close to 0 bps.
In the third demo, prior to transmitting the video, the user requests 80 Kbps be reserved. However, the user's bid is below the required price for the resource, and the request is rejected. Only the web interface is shown below, since the video would not be transmitted in this case.
In the fourth demo, the user requests 160 Kbps and is willing to pay an "acceptable" amount. The policy server accepts the requests, and the background traffic bandwidth allocation is reduced (although 10 Mbps is still being generated). The video quality and rate are very good.
In the fifth demo, the video is finished and the resource is released. The policy server database contents are updated to reflect the newly available resource level.
last updated 1-February-2002