traffic-control: Add additional example documentation
This commit is contained in:
@@ -26,14 +26,51 @@
|
||||
#include "ns3/applications-module.h"
|
||||
#include "ns3/traffic-control-module.h"
|
||||
|
||||
// This simple example shows how to use TrafficControlHelper to install a QueueDisc on a device.
|
||||
// The default QueueDisc is a pfifo_fast with max number of packets equal to 1000 (as in Linux)
|
||||
// This simple example shows how to use TrafficControlHelper to install a
|
||||
// QueueDisc on a device.
|
||||
//
|
||||
// The default QueueDisc is a pfifo_fast with max number of packets equal to
|
||||
// 1000 (as in Linux). However, in this example, we change from the default
|
||||
// to instead use a ns3::RedQueueDisc with a MaxPackets value of 10000.
|
||||
//
|
||||
// Network topology
|
||||
//
|
||||
// 10.1.1.0
|
||||
// n0 -------------- n1
|
||||
// point-to-point
|
||||
//
|
||||
// The output will consist of a number of traced changes to queue lengths
|
||||
// such as:
|
||||
//
|
||||
// DevicePacketsInQueue 0 to 1
|
||||
// TcPacketsInQueue 5 to 4
|
||||
// TcPacketsInQueue 4 to 5
|
||||
// DevicePacketsInQueue 1 to 0
|
||||
//
|
||||
// and an average throughput:
|
||||
//
|
||||
// Average throughput: 8.72854 Mbit/s
|
||||
//
|
||||
// The final output displays the number of drops at the TC layer and the
|
||||
// netdevice layer. These statistics highlight the fact that for
|
||||
// PointToPointNetDevice, the drops at the device layer are actually
|
||||
// requeued at the TC layer, so the true packet drops (39 in this case)
|
||||
// must be traced at the TC layer.
|
||||
//
|
||||
// *** Source stats ***
|
||||
// Number of packets dropped by the TC layer: 39
|
||||
// Number of packets dropped by the netdevice: 3914
|
||||
// Number of packets requeued by the TC layer: 3914
|
||||
// Number of actually lost packets: 39
|
||||
//
|
||||
// If one were to increase the size of the PointToPointNetDevice's
|
||||
// DropTailQueue from 1 to a larger number (e.g. 1000), one would observe
|
||||
// that the number of packets dropped would go to zero, but the latency
|
||||
// and QoS would not be controllable. This is the so-called bufferbloat
|
||||
// problem, and illustrates the importance of having a small device queue
|
||||
// so that the standing queues build in the traffic control layer where
|
||||
// they can be managed by advanced queue discs rather than in the
|
||||
// device layer.
|
||||
|
||||
using namespace ns3;
|
||||
|
||||
|
||||
Reference in New Issue
Block a user