Dear ns users,
I have this simulation that sets up a counter counting incoming packets with payload, and SYN packets (without payload + sequence No.==0) between the HTTP server and client. I used RenoTCP, and WebTraf for generating web traffic. I observe that normally SYN rate equals PP rate (rate of packets with payload) within every observation period. But when I increase the request rate of client, and the server side bandwidth is not large enough, I start to drop SYN packets to a certain rate at the observation point (the gateway of server side), but find that the PP rate is not taken down as low as the SYN rate, it fluctuates between the rates 2~7 times higher than the passed SYN rate. I couldn't figure out why that's the case, is it because of my way of recognizing SYN packets is not right, or because of the way of RenoTCP or WebTraf works? Experienced users, please help me. Thanks in advance. Regards, Linlin Xie
