Impact of background traffic testing with kernel 2.6.18-rc4

From Wiki

Contents 1 Setup 2 Background Traffic 3 Observation 4 Results

Setup

Background Traffic

Traffic type	Description
I	This traffic is generated from the default parameters used in Surge, a Web traffic generator. It uses the body with a lognormal distribution and the tail with a pareto distribution. It consits of 93% body and 7% tail (i.e., 93% flow arrivals follow the lognormal distribution), the shape parameter of the Pareto distribution is 1.1, and the mean and standard deviation of the lognormal distribution are 9.357 and 1.318 respectively, and the minimum file size for the Pareto distribution is 133 KB. THe arrival time of flows follows an exponential distribution with intensity 0.2.
II	Type II adds more rate variance to Type I. It consists of a Web traffic with 70% body and 30% tail. The minimum file size of the Pareto distribution is 1MB. The arrival time of flows follows an exponential distribution with intensity 0.6. The remaining parameter values are the same as those of Type I.
III	The variance in the flow size distribution of this traffic type is between that of Types I and II. It has 10% of all the arriving flows are Peer-to-peer traffic (P2P). The P2P traffic are generated with a Pareto distribution whose shape parameter is 1.1 and the minimum file size is 3MB. The remaining 90% flows are Type I traffic. The inter-arrival time follows an exponential distribution with intensity 0.2. These values are chosen to generate the same amount of traffic as Type I.
IV	This traffic type removes a Pareto distribution from the Web traffic to further reduce the variance of flow sizes from Type I. All flow sizes follow a lognormal distribution with mean 13.357 and standard deviaiton 1.318. These values are chosen to generate the same amount of traffic as Type I.

Observation

Results

• Bandwidth 400Mbps

BK Type	Buffer Size & RTT	#HS flows	CUBIC	HTCP	HSTCP	TCP	Link Utilization
Type I	0.5*BDP & 90ms	2	12345	12345	12345	12345
		4	12345	12345	12345	12345
		8	12345	12345	12345	12345
		16	12345	12345	12345	12345
		32	12345	12345	12345	12345
Type II	0.5*BDP & 90ms	2	12345	12345	12345	12345
		4	12345	12345	12345	12345
		8	12345	12345	12345	12345
		16	12345	12345	12345	12345
		32	12345	12345	12345	12345
Type III	0.5*BDP & 90ms	2	12345	12345	12345	12345
		4	12345	12345	12345	12345
		8	12345	12345	12345	12345
		16	12345	12345	12345	12345
		32	12345	12345	12345	12345
Type IV	0.5*BDP & 90ms	2	12345	12345	12345	12345
		4	12345	12345	12345	12345
		8	12345	12345	12345	12345
		16	12345	12345	12345	12345
		32	12345	12345	12345	12345