Generating Traffic Features

Generating CBR traffic from Set 1 and Set 2, the user can specify the size $L$ of the packets to be transmitted, the total generation time $T$, and the number of bytes for frame $D$. The amount of traffic is specified in terms of frames. Clearly, the number of packets $N$ generated per frame depends of frame size and packet size, that is $N = \frac{D}{L}$.

The generator supports ATM traffic generation only using Set 1. However, IP traffic is supported by all other generation set. In IP option the host name and destination port must be given. In ATM option, VPi, VCi and the Traffic Classes UBR or CBR must be specified, For CBR traffic only the peak rate must be input. For Markov modulated traffic, the model name must be provided.

In Set 1, the packets generated from CBR and Markov Modulated models a frame can be transmitted in two forms: they are either sent one after another at the beginning of the frame generation interval or each packet is sent uniformly spread over the interval. The choice is made using the option Interval between packets. (min indicates that packets are transmitted in sequence at the link rate, as shown in Figure , and max indicates that packets are uniformly spread during the interval between two frames, as indicated in Figure .)

Figure: Generation mode - min

Figure: Generation mode - max

It should be observed in Figure that the generation process sleeps between two consecutive frames. In this case, after transmitting the last packet of a frame, the generator calculates the residual time till the beginning of the transmission of the next frame. If the residual time is positive, then the usleep() system call is executed, otherwise the next frame is transmitted immediately. Note that only one usleep() system call is executed for each frame, and no busy wait is used. The Figure shows the packets being transmitted uniformly over the interval between frames. In this case, the generation process sleeps after transmitting each Packet.

The user can launch a remote traffic receiver, using the receiver options as shown in Figure . The captures the packets sent by the traffic generator and creates a trace file, used to calculate statistics. This module requires the destination host and UDP port.
WARNING: Make sure the firewalls of the sender and receiver machines are open for the chosen UDP port before generating traffic.

The parameters to be given are:

Remote Login

- This parameter specifies the user remote login. The user needs a remote account, as well permition to execute the receiver binary traffgen_recv on remote host.

Local Trace File Name

- The name of trace file that will be generated by the IP Traffic Receiver. The trace collected is sent back to the machine that launched the receiver (and generated traffic), and stored in the file Local Trace File Name, specified in the parameter.

Shell Command

- This specifies the type of remote connection. The user can choose security connection (ssh) or just remote connection ( rsh). Please see the respective man pages for details. The remote machine needs to provide a connection for user's authentication. ssh allows configurations to avoid asking for passwords, see tips below.

Receiver Directory

- The receiver binary files must be copied to directory at the remote account. If it has already been done, the path must be indicated. Otherwise, the TANGRAM2_HOME environment variable will be considered.

Generate compressed trace

- The trace file can be sent compressed from the IP Traffic Receiver to the local machine if this option is selected.

TIPS: The ssh client can perform a login without asking for a password. In the remote user's home create the file .ssh/environment, with the line
PATH=/bin:/usr/bin:/usr/local/bin:$TANGRAM2_HOME. The file .ssh/known_hosts
should contain the remote host name.