Team LiB
Previous Section Next Section

Chapter 7. Congestion Avoidance Through Drop Policies

QoS Exam Objectives

This chapter covers the following exam topics specific to the QoS exam:

  • Describe the drawbacks tail drop as a congestion control mechanism

  • Describe the elements of a RED traffic profile

  • Describe Weighted Random Early Detection and how it can be used to prevent congestion

  • Identify the Cisco IOS commands required to configure and monitor DSCP-based CB-WRED

  • Explain how ECN interacts with WRED in Cisco IOS

Quality of service (QoS) congestion-avoidance tools help prevent congestion before it occurs. These tools monitor queue depth, and before the queue fills, they drop some packets. The computers sending the packets might reduce the frequency of sending packets in reaction to the packet loss, particularly if the application sending the data uses TCP. In the moments after the congestion-avoidance tool discards packets, congestion is reduced, because less traffic is sent into the network.

Cisco congestion-avoidance tools rely on the behavior of TCP to reduce congestion. TCP flows slow down after packet loss. By discarding some TCP packets before congestion gets bad, congestion-avoidance tools may actually reduce the overall number of packets dropped, and reduce congestion, thereby indirectly reducing delay and jitter.

The first section of this chapter begins with a review of TCP and a description of how TCP slows down after packet loss. Following that, several of the problems that can be solved using congestion-avoidance tools are described, namely tail drop, global synchronization, and TCP starvation. The concepts section of this chapter ends with coverage of Random Early Detection (RED), which defines several algorithms, which are the basis of IOS congestion-avoidance tools.

Weighted RED (WRED) and Explicit Congestion Notification (ECN) are two congestion-avoidance tools available in IOS. These are covered in sequence, with the differences between the two highlighted at the end of the chapter.

    Team LiB
    Previous Section Next Section