As far as saturation, I was able to saturate a 100mbit/s network with 2 linux clients using ttcp and netperf netperf: http://www.netperf.org/netperf/NetperfPage.html ttcp: ftp://ftp.sgi.com/sgi/src/ I was using it to test firewalls and NICs. ---- Dani D. Roisman [EMAIL PROTECTED] > -----Original Message----- > From: Hariharan L Thantry [SMTP:[EMAIL PROTECTED]] > Sent: Thursday, June 29, 2000 8:44 PM > To: unlisted-recipients > Cc: [EMAIL PROTECTED] > Subject: Some pointers needed > > Hi, > > I need to do some kind of performance evaluation of TCP's various timer > back-off strategies, congestion avoidance schemes, recovery process in > sender etc. I currently have a network > with two small subnets set up. One subnet having two machines (clients), > and the other a single server machine. I use one machine to act as a > gateway to switch packets between these two networks. > > Here's the problem. My raw bandwidth is 10MBps and there seems to be no > hope to actually saturate that kind of bandwidth with traffic generated > by > just two clients. And to do a study on TCP behaviour during congestion, I > would need to make sure that some packets are dropped. So, I think the > best strategy is to make the gateway drop the packets. Another option is > to have some dynamic traffic shaping scheme on the interface card on the > server-side subnet which would allow me to vary the observed bandwidth on > the server side and hence make the server drop some packets. I am not too > sure if this would work. > > If anybody has worked on this kind of stuff, please let me know. I'd like > to know exactly what I need to do to get running on this. Any pointers > you may have to share would be great too!! > > Thanks > Hari > > - > To unsubscribe from this list: send the line "unsubscribe linux-net" in > the body of a message to [EMAIL PROTECTED] - To unsubscribe from this list: send the line "unsubscribe linux-net" in the body of a message to [EMAIL PROTECTED]
