CCVP: Quality of Service (QoS) Basics – Part 3

Now that we've discussed some of the basic concepts and configuration fundamentals for topics found in Cisco's QoS course, let's shift our focus to a few practice exam questions. By the way, these are not "brain dump" questions. These are questions that I've written that reflect the concepts tested on the QoS exam.

Following is a series of five QoS exam-style questions. After answering each of the questions, you can then check your answers and review the detailed explanations. Here we go, and good luck!

Questions

Question 1:

Which statement is true regarding the following configuration?

class-map C_VOICE

 match protocol rtp audio

class-map C_VIDEO

 match protocol rtp video

class-map C_TRANSACTIONAL

 match dscp af11 af12 af13 cs1

!

policy-map P_MQC

 class C_VOICE

  priority percent 15

 class C_VIDEO

  priority percent 35

 class C_TRANSACTIONAL

  bandwidth percent 20

 class class-default

  fair-queue

!

interface serial 0/0

 service-policy output P_MQC

1.            Traffic marked with an IP Precedence value of 1 will be placed in the class-default class.

2.            Packets within the priority queue will be queued using FIFO, while packets within the class-default class will be queued using WFQ.

3.            Voice and video packets will be placed in separate queues.

4.            The configuration is not valid, because an LLQ configuration such as this can only be applied in the inbound direction.

Question 2:

When sending traffic across a service provider network, using BGP as the exterior gateway routing protocol, the service provider can use BGP attributes (e.g. a community string or an autonomous system path) to cause edge routers to mark traffic with IP Precedence or QoS group values. Identify this QoS mechanism.

1.            CEF

2.            QBBP

3.            MIB

4.            Pre-Classify

Question 3:

Which of the following statements are true concerning QoS configuration on a Cisco Catalyst 2950 Series switch? (Choose 3)

1.            When a port trusts CoS values, frames with a CoS of 5 are placed in Queue #3 by default.

2.            The default CoS-to-DSCP mapping for a CoS of 5 is to a DSCP value of 40.

3.            The expedite queue is enabled by default, but it only contains traffic for routing traffic, which has a CoS value of 6.

4.            Only Queue #4 can be configured as an expedite queue.

Question 4:

Identify two of the following statements which are correct regarding WRED. (Choose 2)

1.            WRED is a congestion management mechanism.

2.            WRED uses an interface's average queue depth when determining if a packet should be dropped.

3.            WRED uses a packet's IP Precedence or DSCP marking when determining if a packet should be dropped.

4.            WRED's Mark Probability Denominator (MPD) parameter equals the probably of discard when the queue depth equals the RED profile's maximum threshold value.

Question 5:

Consider the following configuration:

class-map TRANSACTIONAL

 match protocol http

!

policy-map CBSHAPING

 class TRANSACTIONAL

  bandwidth 64

  shape average 128000

  shape adaptive 64000

!

interface serial 0/1

 service-policy input CBSHAPING

Why is the configuration invalid?

1.            The "bandwidth" command should be "bandwidth 64000" instead of bandwidth "64."

2.            CB-Shaping and CB-WFQ cannot be configured in the same policy-map.

3.            A policy-map containing a shaping policy cannot be applied to an interface in the inbound direction.

4.            The "shape adaptive" command is not valid in the absence of a FECN adaptation configuration.

Answers

Question #1 Answer: B

Traffic marked with an IP Precedence of 1 has the six left-most bits of its Type of Service (ToS) byte set to 001000, which identical to the six left-most bits of the ToS byte for a Class Selector 1 (CS1) marking. In fact, the Class Selector category of DSCP Per Hop Behaviors (PHBs) provides pure backwards compatibility with IP Precedence markings. Therefore, traffic marked with an IP Precedence value of 1 will be placed in the C_TRANSACTIONAL class.

In an MQC configuration, with the exception of the Cisco 7x00 Series router platform, only the class-default queue can be configured to treat traffic within that queue using the Weighted Fair Queuing (WFQ) algorithm. All other queues, even the priority queue, can only use First-In First-Out (FIFO) queuing. Therefore, in this configuration, the voice and video packets in the priority queue are treated with FIFO queuing within the queue, while packets in the class-default class' queue are treated with WFQ within the queue, due to the fair-queue command issued in policy-map-class configuration mode for the class-default class.

Also, even though two class-maps use the priority keyword, to give traffic within those queues priority treatment, there is only a single priority queue in a LLQ configuration. Therefore, the voice and video packets share the same priority queue.

Finally, MQC queuing configurations can only be applied to an interface in the outbound direction. Therefore, the service-policy output P_MQC command shown in this configuration is valid.

Question #2 Answer: B

QoS Policy Propagation through BGP (QPPB) is a legacy approach for classification and marking traffic at the edge of a BGP network. Although BGP itself is not performing any QoS classification, an administrator can configure QPPB to mark traffic based on BGP attributes such as a BGP community string or an autonomous system (AS) path. Since QPPB is a legacy approach for classification and marking, only IP Precedence and QoS group markings are supported, while DSCP markings are not supported. Also, note that Cisco Express Forwarding (CEF) must be enabled on a router prior to QPPB configuration. CEF enhances packet forwarding by storing Layer 2 and Layer 3 forwarding information in memory. The table containing the Layer 2 information is called the Adjacency Database, and the table containing the Layer 3 information is the Forwarding Information Base (FIB).

Question #3 Answer: A, B, D

Some default QoS settings on a Cisco Catalyst 2950 Series switch are not optimal for VoIP traffic. For example, a Catalyst 2950 has four queues. However, no expedite (i.e. priority) queue is configured by default. Also, if a port is trusting CoS values, the default queue mappings place CoS 5 (e.g. VoIP) traffic into Queue #3, while only Queue #4 can be configured as an expedite queue. A Catalyst 2950 also has the ability to take traffic marked with a CoS value and remark that traffic with a DSCP value. Unfortunately, the Layer 3 DSCP marking coming from a Cisco IP Phone will be marked down with the default CoS-to-DSCP mapping.

The Cisco IP Phone sends voice traffic, by default, marked with a Layer 2 CoS marking of 5 and a Layer 3 DSCP marking of EF (i.e. Expedited Forwarding), which has a decimal equivalent value of 46. However, the default CoS-to-DSCP mapping on a Catalyst 2950 will examine the traffic coming from the Cisco IP Phone and remark the traffic with a DSCP value of a 40, because of the CoS-to-DSCP marking table which specifies that a CoS value of 5 is remarked at Layer 3 with a DSCP value of 40.

Question #4 Answer: B, C

Weighted Random Early Detection (WRED) is a congestion avoidance mechanism. Different IP Precedence and DSCP values each have RED profiles, which determine the probability of packet discard based on an interface's average queue depth. A RED profile specifies the minimum threshold, maximum threshold, and probability of discard when the average queue depth equals the maximum threshold. When the average queue depth is at or below the minimum threshold, there is zero probability of discard (i.e. the "no drop" mode). When the average queue depth is greater than the maximum threshold, there is a 100 percent probability of discard (i.e. the "full drop" mode). When the average queue depth is greater than the minimum threshold and equal to or less than the maximum threshold, there is a possibility of packet discard (i.e. the "random drop" mode). The probably of discard when the average queue depth equals the maximum threshold equals 1/MPD, where MPD is the configurable Mark Probability Denominator. For example, if the MPD equals 4, the probably of discard when the average queue depth equals the maximum threshold is 1/4 (i.e. 25 percent).

Question #5 Answer: C

The class-map and policy-map portions of the configuration are valid. Specifically, the class-map is matching HTTP traffic, and the policy-map is configured for both CB-Shaping and CB-WFQ. The CB-Shaping configuration is using a CIR of 128 kbps, and the use of the "average" keyword, as opposed to the "peak" keyword, indicates that no more than the Bc number of bits will be sent per timing interval.

NOTE: The unit of measure for the CIR in the "shape" command is bps, not kbps.

The shape adaptive 64000 command indicates that in the presence of Backwards Explicit Congestion Notification (BECN) messages being received inside of frames coming from the service provider, the CIR can be reduced at a rate of 25 percent per timing interval until a minimum bandwidth of 64 kbps is reached. Although not shown in this configuration, the shape fecn-adapt command in policy-map-class configuration mode allows a receiving router to respond to a frame, from a service provider, with the Forward Explicit Congestion Notification (FECN) bit set. The response to such a frame is for the router to generate a Q.922 test frame destined for the transmitting router. As this test frame passes through the service provider's cloud, the service provider can mark the frame's BECN bit, and when the transmitting router receives this frame, it can reduce its CIR by 25 percent as previously described. This approach is useful when most of the traffic between the two routers is flowing in one direction. That is, there is little opportunity for the "receiving" router to transmit frames to the "transmitting" router, which the service provider could potentially mark with a BECN bit.

In addition to CB-Shaping, this configuration also performs CB-WFQ, which guarantees a minimum amount of bandwidth of 64 kbps for HTTP traffic, if HTTP traffic needs that much bandwidth. However, not all QoS mechanisms can be applied in the inbound direction. For example, while CB-Policing is supported for traffic coming into or going out of an interface, CB-Shaping can only be applied to traffic exiting an interface.


Well, how did you do? Remember, while these are not exact questions from the QoS exam, these questions reflect the style and difficulty of questions found on the exam. If you would like some additional practice, click HERE to download a .PDF containing additional practice QoS exam questions. Or, visit www.voipcertprep.com to purchase the full 51-question practice exam along with 12 video demonstrations. Sorry if that sounds a bit markety, but these are study aids that I've personally created, and I'm sure they can help anyone seriously preparing for the QoS exam.

My time is running short, since I'm only blogging through the month of December, and based on your requests at the  beginning of the month we still have a couple of major topics to cover. Specifically, we still want to talk about setting up a home lab and also give you a glimpse into the CCNA Voice concentration. So, coming up next time, I'll be sharing some tips for setting up a home lab for practicing VoIP configs.

Hope you're enjoying the holiday season!

Kevin    

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2008 IDG Communications, Inc.