NPR Tutorial >> Filters, Queues and Bandwidth | TOC |
In the dumbbell example that we have been developing, we have been displaying the traffic bandwidth along the paths taken by the two flows from n1p2 and n1p3. This page shows you how you can also monitor queue length and the number of packet drops at the bottleneck port.
Recall that traffic from n1p2 and n1p3 are directed to output queue 64 at port 1.4 using packet filters. That queue has a queueing capacity of 1,500,000 bytes (i.e., its threshold), and the oututp port has an output capacity of 300 Mbps. The charts will show, as expected, that queue 64 is full during the contention period [618, 630], and that the queue is empty otherwise. Also, the packet drops do occur only during the contention period.
An example based on some of the concepts discussed on this page is given in NPR Tutorial => Examples => Filters, Queues and Bandwidth. The sections in this page are:
We use the Monitoring => ReadQLength menu item in the port menu to monitor the length of a queue. Here is a recipe for monitoring the length of queue 64.
The queue length chart will appear.
The Add Parameter dialogue box will appear.
The Queue Length chart will appear with the line labeled as ReadQLength. Accept the 1-second polling rate, and monitor the absolute value of the queue length counter.
The queue length chart is shown on the left, and the bandwidth chart is shown on the right for comparison. (If the two charts do not appear side-by-side, widen your browser window until the are.) As expected, the queue length (of queue 64) is equal to the 1.5 MB threshold when both flows are active.
Also, we can verify that it will take much less than one second
(the chart polling interval) for queue 64 to overflow
once the second flow starts.
The excess input rate is 100 Mbps since the aggregate input rate is
400 Mbps, and the bottleneck output rate is 300 Mbps.
So, we expect that it should only take about 0.12 seconds
(12 Mb/100 Mbps) to overflow the queue.
We use the Monitoring => ReadRegisterPacket menu item at the center of the NPR icon to monitor the number of packets dropped by the Queue Manager at port 1.4. Recall that register pair 31 counts the number of bytes and the number of packets dropped by the Queue Manager due to a queue overflow. Here is a recipe for monitoring the number of drops:
The QM Drops chart will appear.
The Add Parameter dialogue box will appear.
The QM Drop chart appears with the line labeled as ReadRegisterPacket.
We will change the name of the line from ReadRegisterPacket to QM Drops.
This dialogue box indicates rate or absolute number of packets being dropped by the Queue Manager. (We could have selected rate in the Add Parameter dialogue box earlier.)
The packet drop chart is shown on the left, and the bandwidth chart is shown on the right for comparison. As expected, the packet drops occur only during the contention period when both flows are active.
Revised: Tue, Aug 26, 2008
NPR Tutorial >> Filters, Queues and Bandwidth | TOC |