BURST TEST

Scenario:

 The Burst scenario is as following:
        - class 2:0, 3:0, 6:0 backlogged during the whole test.
        - 1 Burst for class 4:0 
        - 2 Double Simultaneous Bursts for class 4:0 and 5:0 

The following is a screen capture of the monitoring tool to show the position of the bursts, as the plots overlap, you cannot see all the colors.

HTB Graphical Results

The following is a screen capture of the monitoring tool during HTB tests

The following represents dequeues exploited with gnuplot, the x unit is the second (HTB).

The following represents an other occurence of the same test, but with a different time-unit, the x unit is the 100th of milli-second. The colors are not the same as the monitor, you can identify which color belongs to which class by yourself (HTB).

The following represents the sum of the dequeues of all the classes 2:0, 3:0, 4:0, 5:0, 6:0 (HTB).

The following represents the sum of the dequeues of the classes 2:0, 3:0, 4:0, 5:0 (HTB).

The following represents the delays of all the classes temporal resolution: 1 sec (HTB).

The following represents the delays of all the classes temporal resolution: 100 msec. The burst represented is the double burst. (HTB)

HFSC Graphical Results

The following is a screen capture of the monitoring tool during HFSC tests

The following represents the same data exploited with gnuplot, the x unit is the second. (HFSC)

The following represents an other occurence of the same test, but with a different time-unit, the x unit is the 100th of milli-second. (HFSC)

The following represents the sum of the dequeues of all the classes 2:0, 3:0, 4:0, 5:0, 6:0. (HFSC)

The following represents the sum of the dequeues of the classes 2:0, 2:3, 2:4, 2:5. (HFSC)

The following represents the delays of all the classes temporal resolution: 1 sec. (HFSC)

The following represents the delays of all the classes temporal resolution: 100 msec. The burst represented is the double burst. (HFSC)

BACK TO HTB

After I made all those mesurements, I was definitly not happy with the fact that the sum of classes 2:0, 3:0, 4:0 and 5:0 was not constant for the HTB qdisc, so I created the HTB2 configuration in which a qdisc is inserted under the 1:2 class, see under.

if [ $1 == htb ]; then
  $TC qdisc add dev $DEV root handle 7: htb
  $TC class add dev $DEV parent 7:0 classid 7:1 htb rate 500000 ceil 500000
  $TC qdisc add dev $DEV parent 7:1 handle 1: htb
  $TC class add dev $DEV parent 1:0 classid 1:1 htb rate 500000 ceil 500000
  $TC class add dev $DEV parent 1:1 classid 1:2 htb prio 4 rate 200000 ceil 500000
  $TC class add dev $DEV parent 1:1 classid 1:3 htb prio 4 rate 300000 ceil 500000
  $TC class add dev $DEV parent 1:2 classid 1:22 htb prio 4 rate 150000 ceil 500000
  $TC class add dev $DEV parent 1:2 classid 1:23 htb prio 4 rate 50000 ceil 500000
  $TC class add dev $DEV parent 1:2 classid 1:24 htb prio 2 rate 120000 ceil 120000
  $TC class add dev $DEV parent 1:2 classid 1:25 htb prio 2 rate 80000 ceil 80000
else
  if [ $1 == htb2 ]; then
    $TC qdisc add dev $DEV root handle 7: htb
    $TC class add dev $DEV parent 7:0 classid 7:1 htb rate 500000 ceil 500000
    $TC qdisc add dev $DEV parent 7:1 handle 1: htb
    $TC class add dev $DEV parent 1:0 classid 1:1 htb rate 500000 ceil 500000
    $TC class add dev $DEV parent 1:1 classid 1:2 htb prio 4 rate 200000 ceil 500000
    $TC class add dev $DEV parent 1:1 classid 1:3 htb prio 4 rate 300000 ceil 500000
    $TC qdisc add dev $DEV parent 1:2 handle 10: htb
    $TC class add dev $DEV parent 10:0 classid 10:2 htb prio 4 rate 500000 ceil 500000
    $TC class add dev $DEV parent 10:2 classid 10:22 htb prio 4 rate 150000 ceil 500000
    $TC class add dev $DEV parent 10:2 classid 10:23 htb prio 4 rate 50000 ceil 500000
    $TC class add dev $DEV parent 10:2 classid 10:24 htb prio 2 rate 120000 ceil 120000
    $TC class add dev $DEV parent 10:2 classid 10:25 htb prio 2 rate 80000 ceil 80000
  else
    $TC qdisc add dev $DEV root handle 1: hfsc

For the whole config file, download the monitor tool.

HTB Other Config Graphical Results

All the results with this new configuration are similar to the previous HTB configuration except for the 3 following plots, those show that the new config solves the problem.

The following represents the dequeues for HTB 2 configuration, as you can see, the bursts have no impact on classe 6:0. (HTB2)

The following represents the sum of the dequeues of the classes 2:0, 3:0, 4:0, 5:0, the sum is constant. (HTB2)

The following represents the delays of all the classes this shows where the time to transmit the bursts came from, not from 6:0 but from the others. (HTB2)