<div>Thanks for the pointer. </div>
<div> </div>
<div>But why fairness and friendliness are equally important?</div>
<div> </div>
<div>I though all those congestion algorithms are already intra-protocol fair (otherwise, it will be a design flaw and won't get implemented)</div>
<div> </div>
<div>As to inter-protocol friendliness, looks to me, every TCP congestion controls are already friendly enough comparing with UDP.<br></div>
<div>-Adam<br></div>
<div class="gmail_quote">On Tue, Jul 8, 2008 at 11:40 PM, Lars Eggert <<a href="mailto:lars.eggert@nokia.com">lars.eggert@nokia.com</a>> wrote:<br>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">Hi,
<div class="Ih2E3d"><br><br>On 2008-7-8, at 20:26, ext adam maxiaodong wrote:<br>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">In my test, vegas doesn't perform better than others, but I guess vegas will<br>perform better when network path has long queuing. hope this help!<br>
</blockquote><br></div>if you haven't done so, I highly recommend Alejandro and you take a look at RFC 5166 ("Metrics for the Evaluation of Congestion Control Mechanisms").<br><br>Throughput is one important metric to look at (Section 2.1.1), but there are many others that are equally important. Picking a congestion control algorithm only on the basis of achievable throughput is missing the bigger issues, such as fairness.<br>
<font color="#888888"><br>Lars<br><br></font></blockquote></div><br>