Traffic / DUT profile

Devices Under Test (DUTs)

Traffic Theory

The test matrix and seeps

Each traffic profile is an N dimentional rectangular matrix. A simple example would be the two dimentional matrix of Number of clients say 1-11 vs the transfer sizs, say 0 kByte to 1 Mbyte. Each axis, or sweep, is given a number of steps. At the intersection of each sweeps steps we have an argument vector that the tests are run with.

 
Numer of clients [1-11] in 6 steps
Transfer size [0 - 1Mbyte] in 6 steps 1 3 5 7 9 11
0            
200,000            
400,000            
600,000            
800,000            
1,000,000            

The test driver then walks the matrix, setting knobs to there specified values for the cell in question and waiting a proscribed amount of time to complete data accuisition for this cell. The above exmaple uses sweeps that are numeric and linearly interpolated. There are many other options that are possible. Review the documentation to find out more.

Knobs

Knobs are the way things get done. The software is based on the analogy of knobs and widgets. A knob is the input side and the widgets respond to the knobs commands and act accordingly.

for example a simple knob is a location for example 'knobspace', and a name for example 'mtu' or 'avgfilesize'.

Then there are smart knobs. Smart knobs allow one to build and orchistrate relationships between different knobs. For example, if you'd like to set 5 knobs to identical values a 'samesame' smart knob would be used. If you would like the data value to be spread around several knbs then the 'roundrobin' smart knob would be used. For other examples and an expanded explenation see the documentation.

The Test Matricies

  • Study null
    Sweep # datatype scale/interpolation start end # steps Knobs
    1 int linear 64 0 1
    Cell length : 60 Cell cnt : 1 aprox runtime (61) seconds per DUT per config

  • Study pkthoze_size
    Sweep # datatype scale/interpolation start end # steps Knobs
    1 int linear 1 1 1 knobspace/samesame(top_beakon_0,bottom_beakon_0,top_beakon_1,bottom_beakon_1)|processes
    2 int linear 1 4 4 knobspace/roundrobin(top_pkthoze_0,bottom_pkthoze_0,top_pkthoze_1,bottom_pkthoze_1)|processes
    3 int linear 30 1450 8 knobspace/samesame(top_pkthoze_0,bottom_pkthoze_0,top_pkthoze_1,bottom_pkthoze_1)|pktsize
    4 int linear 1 5 2 knobspace/samesame(top_pkthoze_0,bottom_pkthoze_0,top_pkthoze_1,bottom_pkthoze_1)|numpkts
    Cell length : 60 Cell cnt : 64 aprox runtime (4690) seconds per DUT per config

  • Study wget_avgsfilesize
    Sweep # datatype scale/interpolation start end # steps Knobs
    1 int linear 64 0 1 knobspace/roundrobin(top_wget_0,bottom_wget_0,top_wget_1,bottom_wget_1)|processes
    2 int log 1500 70 1 knobspace/samesame(top_wget_0,bottom_wget_0,top_wget_1,bottom_wget_1)|mtu
    3 int linear 0 300000 10 knobspace/samesame(top_wget_0,bottom_wget_0,top_wget_1,bottom_wget_1)|avgfilesize
    Cell length : 60 Cell cnt : 10 aprox runtime (725) seconds per DUT per config

  • Study wget_mtuslide
    Sweep # datatype scale/interpolation start end # steps Knobs
    1 int linear 64 0 1 knobspace/roundrobin(top_wget_0,bottom_wget_0,top_wget_1,bottom_wget_1)|processes
    2 int log 1 300000 5 knobspace/samesame(top_wget_0,bottom_wget_0,top_wget_1,bottom_wget_1)|avgfilesize
    3 int linear 1500 70 10 knobspace/samesame(top_wget_0,bottom_wget_0,top_wget_1,bottom_wget_1)|mtu
    Cell length : 60 Cell cnt : 50 aprox runtime (3285) seconds per DUT per config

    Yes, this page was computer generated. Please report any bugs you find.