QoS and Flow-based

Redirection Configuration Content

Content

CHAPTER 1 QOS CONFIGURATION ....................................... 1-1

1.1 INTRODUCTION TO QOS .......................................................................1-1

1.1.1 QoS Terms ...............................................................................................1-1

1.1.2 QoS Implementation ...............................................................................1-2
1.1.3 Basic QoS Model .....................................................................................1-3

1.2 QOS CONFIGURATION TASK LIST .........................................................1-8

1.3 QOS EXAMPLE ................................................................................. 1-12

1.4 QOS TROUBLESHOOTING .................................................................. 1-15

CHAPTER 2 FLOW-BASED REDIRECTION............................. 2-1

2.1 INTRODUCTION TO FLOW-BASED REDIRECTION ......................................2-1

2.2 FLOW-BASED REDIRECTION CONFIGURATION TASK SEQUENCE ..............2-1

2.3 FLOW-BASED REDIRECTION EXAMPLES ................................................2-2

2.4 FLOW-BASED REDIRECTION TROUBLESHOOTING HELP ..........................2-3

CHAPTER 3 EGRESS QOS CONFIGURATION ........................ 3-1

3.1 INTRODUCTION TO EGRESS QOS ..........................................................3-1

3.1.1 Egress QOS Terms..................................................................................3-1

3.1.2 Basic Egress QoS Model........................................................................3-1

3.2 EGRESS QOS CONFIGURATION ............................................................3-3

3.3 EGRESS QOS EXAMPLES ....................................................................3-7

3.4 EGRESS QOS TROUBLESHOOTING HELP ..............................................3-9

CHAPTER 4 FLEXIBLE QINQ CONFIGURATION .................... 4-1

4.1 INTRODUCTION TO FLEXIBLE QINQ .......................................................4-1

4.1.1 QinQ Technique .......................................................................................4-1

4.1.2 Basic QinQ ...............................................................................................4-1
4.1.3 Flexible QinQ ...........................................................................................4-1

1
QoS and Flow-based
Redirection Configuration Content

4.2 FLEXIBLE QINQ CONFIGURATION TASK LIST .........................................4-1

4.3 FLEXIBLE QINQ EXAMPLE ...................................................................4-3

4.4 FLEXIBLE QINQ TROUBLESHOOTING ....................................................4-5

2
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

Chapter 1 QoS Configuration

1.1 Introduction to QoS

QoS (Quality of Service) is a set of capabilities that allow you to create differentiated
services for network traffic, thereby providing better service for selected network traffic.
QoS is a guarantee for service quality of consistent and predictable data transfer service
to fulfill program requirements. QoS cannot generate extra bandwidth but provides more
effective bandwidth management according to the application requirement and network
management policy.

1.1.1 QoS Terms

QoS: Quality of Service, provides a guarantee for service quality of consistent and
predictable data transfer service to fulfill program requirements. QoS cannot generate
new bandwidth but provides more effective bandwidth management according to the
application requirement and network management.
QoS Domain: QoS Domain supports QoS devices to form a net-topology that provides
Quality of Service, so this topology is defined as QoS Domain.
CoS: Class of Service, the classification information carried by Layer 2 802.1Q frames,
taking 3 bits of the Tag field in frame header, is called user priority level in the range of 0 to
7.

Fig 1-1 CoS priority

ToS: Type of Service, a one-byte field carried in Layer 3 IPv4 packet header to symbolize
the service type of IP packets. Among ToS field can be IP Precedence value or DSCP
value.

1-1
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

Fig 1-2 ToS priority

IP Precedence: IP priority. Classification information carried in Layer 3 IP packet header,
occupying 3 bits, in the range of 0 to 7.
DSCP: Differentiated Services Code Point, classification information carried in Layer 3 IP
packet header, occupying 6 bits, in the range of 0 to 63, and is downward compatible with
IP Precedence.
Internal Priority: The internal priority setting of the switch chip, it’s valid range relates
with the chip, it’s shortening is Int-Prio or IntP.
Drop Precedence: When processing the packets, firstly drop the packets with the bigger
drop precedence, the ranging is 0-1. It’s shortening is Drop-Prec or DP.
Classification: The entry action of QoS, classifying packet traffic according to the
classification information carried in the packet and ACLs.
Policing: Ingress action of QoS that lays down the policing policy and manages the
classified packets.
Remark: Ingress action of QoS, perform allowing, degrading or discarding operations to
packets according to the policing policies.
Scheduling: QoS egress action. Add the packets to the corresponding egress queue
according to the internal priority. And then decide sending and dropping according to Drop
Precedence, sending algorithm and queue weight of egress queue.

1.1.2 QoS Implementation

To implement the switch software QoS, a general, mature reference model should be
given. QoS can not create new bandwidth, but can maximize the adjustment and
configuration for the current bandwidth resource. Fully implemented QoS can achieve
complete management over the network traffic. The following is as accurate as possible a
description of QoS.
The data transfer specifications of IP cover only addresses and services of source
and destination, and ensure correct packet transmission using OSI layer 4 or above
protocols such as TCP. However, rather than provide a mechanism for providing and
protecting packet transmission bandwidth, IP provide bandwidth service by the best effort.
This is acceptable for services like Mail and FTP, but for increasing multimedia business
data and e-business data transmission, this best effort method cannot satisfy the

1-2
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

bandwidth and low-lag requirement.

Based on differentiated service, QoS specifies a priority for each packet at the
ingress. The classification information is carried in Layer 3 IP packet header or Layer 2
802.1Q frame header. QoS provides same service to packets of the same priority, while
offers different operations for packets of different priority. QoS-enabled switch or router
can provide different bandwidth according to the packet classification information, and can
remark on the classification information according to the policing policies configured, and
may discard some low priority packets in case of bandwidth shortage.
If devices of each hop in a network support differentiated service, an end-to-end QoS
solution can be created. QoS configuration is flexible, the complexity or simplicity depends
on the network topology and devices and analysis to incoming/outgoing traffic.

1.1.3 Basic QoS Model

The basic QoS consists of four parts: Classification, Policing, Remark and Scheduling,
where classification, policing and remark are sequential ingress actions, and Queuing and
Scheduling are QoS egress actions.

Fig 1-3 Basic QoS Model

Classification: Classify traffic according to packet classification information and generate
internal priority based the classification information. For different packet types,
classification is performed differently; the flowchart below explains this in detail.

1-3
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

Start

N tag packet

L2 COS value
L2 COS value of the
obtained by the
packet is its own L2
packet as the default
COS
COS(*1)

Trust DSCP Y IP packet

(*2)

N
N

Trust COS Y
N
(*2)

N tag packet

Y
Set Int-Prio as the
DSCP-to-Int-Prio
default ingress Int-
COS -to-Int-Prio conversion according to
Prio
conversion according to DSCP value of the packet
L2 COS value of the
packet

Enter the policing flow

Fig 1-4 Classification process

Note 1: L2 CoS value is considered a property of the packets, there is no relation with the
internal priority obtained of the following flow.

1-4
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

Note 2: Allow Trust DSCP and Trust COS to be configured at the same time, the priority is
as follows: DSCP>COS.
Policing and remark: Each packet in classified ingress traffic is assigned an internal
priority value, and can be policed and remarked.
Policing can be performed based on the flow to configure different policies that
allocate bandwidth to classified traffic, the assigned bandwidth policy may be single
bucket dual color or dual bucket three color. The traffic, will be assigned with different
color, can be discarded or passed, for the passed packets, add the remarking action.
Remarking uses a new Int-Prio value of lower priority to replace the original higher level
Int-Prio value in the packet. COS and DSCP fields will be modifed according to the new
Int-Prio at the egress. The following flowchart describes the operations.

1-5
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

Start

No Whether configure
the policy

Yes

Unrelated action Drop

with the color

Pass

The option is as follows:

Set Int-Prio: Set the internal priority of
the packets

Decide the packet color

and action according to
the policing policy

The specific
Drop
color action

Pass

Select one option of the following:

Set Int-Prio: Set the internal priority of the
packets(*1)
Policied-IntP-Transmit: Drop the internal priority of
the packets(*2)

Drop the
Enter scheduling
packets

Fig 1-5 Policing and Remarking process

Note 1: Int-Prio will be covered with the after setting, Set Int-Prio of the specific color
action will cover Set Int-Prio of the unrelated action with the color.
Note 2: Drop the internal priority of the packets according to IntP-to-IntP map. Source
Int-Prio means to the obtainable Int-Prio in Classification flow or Int-Prio set by the
unrelated action with the color.

1-6
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

Queuing and scheduling: There are the internal priority for the egress packets, the
scheduling operation assigns the packets to different priority queues according to the
internal priority, and then forward the packets according to the priority queue weight and
the drop precedence. The following flowchart describes the scheduling operation.

Start

Remark DSCP and L2 COS fields

of the packets according to
Int-Prio-to-DSCP
Int-Prio-to-COS mapping(*1)

Select the queue according to

IntPrio-to-Queue mapping
Obtain the packet Drop-Prec
according to IntPrio-to-Drop-
Prec

Read the buffer value according

to the queue management
algorithm(WDRR/SP), the drop
precedence and the egress
queue

buffer is
available No

Yes

Place the packets into the

specified queue, and forward the
packets according to the weight
priority

Drop the
Finish packets

Fig 1-6 Queuing and Scheduling process

Note 1: The ingress configures pass-through-cos, pass-through-dscp to forbid the
rewrite of L2 CoS priority and dscp value. At the egress, obtain L2 CoS priority and dscp
value according to the final Int-Prio of the packets, decide whether rewrite L2 CoS priority
and dscp value according to pass-through-cos, pass-through-dscp.

1-7
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

1.2 QoS Configuration Task List

Configure class map

Set up a classification rule according to ACL, CoS, VLAN ID, IPv4 Precedent, DSCP,
IPV6 FL to classify the data stream. Different classes of data streams will be
processed with different policies.
Configure a policy map
After data steam classification, a policy map can be created to associate with the class
map created earlier and enter class mode. Then different policies (such as bandwidth
limit, priority degrading assigning new DSCP value) can be applied to different data
streams. You can also define a policy set that can be use in a policy map by several
classes.
Apply QoS to the ports or the VLAN interfaces
Configure the trust mode for ports or bind policies to ports. A policy will only take effect
on a port when it is bound to that port.
The policy may be bound to the specific VLAN.
It is not recommended to synchronously use policy map on VLAN and its port, or else
the policy map priority of the port is higher.
Configure queue management algorithm
Configure queue management algorithm, such as sp, wdrr, and so on.
Configure QoS mapping
Configure the mapping from CoS to IntP, DSCP to IntP, IntP to DSCP, COS, IntP,
DP or queue.

1. Configure class map.

Command Explanation
Global Mode
Create a class map and enter class
class-map <class-map-name> map mode; the “no class-map
no class-map <class-map-name> <class-map-name>” command
deletes the specified class map.
match {access-group <acl-index-or-name>
| ip dscp <dscp-list>| ip precedence Set matching criterion (classify data
<ip-precedence-list>| ipv6 access-group stream by ACL, CoS, VLAN ID, IPv4
<acl-index-or-name> | ipv6 dscp Precedent, IPv6 FL or DSCP, etc) for
<dscp-list>| ipv6 flowlabel the class map; the no command
<flowlabel-list>|vlan <vlan-list> | cos deletes specified matching criterion.
<cos-list>| vlan range <vlan-list>}

1-8
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

no match {access-group | ip dscp | ip

precedence | ipv6 access-group | ipv6
dscp | ipv6 flowlabel | vlan | cos| vlan
range}

2. Configure a policy map

Command Explanation
Global Mode
Create a policy map and enter policy
policy-map <policy-map-name>
map mode; the no command deletes
no policy-map <policy-map-name>
the specified policy map.
After a policy map is created, it can be
associated to a class. Different policy
class <class-map-name> [insert-before
or new DSCP value can be applied to
<class-map-name>]
different data streams in class mode;
no class <class-map-name>
the no command deletes the specified
class.
set internal priority <new-intp> Assign a new internal priority for the
no set internal priority classified traffic; the no command
cancels the new assigned value.
Single bucket mode: Configure a policy for the classified
policy <bits_per_second> flow. The non-aggregation policy
<normal_burst_bytes> ({exceed-action command supports three colors.
ACTION} ) Analyze the working mode of the
Dual bucket mode: token bucket, whether it is singe rate
policy <bits_per_second> single bucket, single rate dual bucket,
<normal_burst_bytes> [pir dual rate dual bucket, set
<peak_rate_bps>] | corresponding action to different color
<maximum_burst_bytes> [{exceed-action packets. The no command will delete
ACTION | violate-action ACTION }] the mode configuration.
ACTION definition:
drop | transmit | set-internal-priority
<intp_value> | policied-intp-transmit
no policy
Apply a policy to classified traffic; the
policy aggregate <aggregate-policy-name>
no command deletes the specified
no policy aggregate
policy set.
accounting Set statistic function for the classified

1-9
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

no accounting traffic. After enable this function under

the policy class map mode, add
statistic function to the traffic of the
policy class map. In single bucket
mode, the messages can only red or
green when passing policy. In the print
information, there are two
colors(green and red) of the packets.
In dual bucket mode, there are three
colors(green, red and yellow) of the
packets.
Policy class map configuration mode
drop Drop or transmit the traffic that match
no drop the class, the no command cancels
the assigned action.
transmit
no transmit

3. Apply QoS to port or VLAN interface

Command Explanation
Interface Configuration Mode
mls qos trust {cos | dscp} Configure port trust; the no command
no mls qos trust {cos | dscp} disables the current trust status of the
port.
Configure the default CoS value of the
mls qos cos {<default-cos>}
port; the no command restores the
no mls qos cos
default setting.
Configure the default internal priority
mls qos internal-priority {<default-intp> }
value of the port, the no command
no mls qos internal-priority
restores the default setting.
service-policy input <policy-map-name> Apply a policy map on the port, the no
no service-policy input command deletes the specified policy
{<policy-map-name>} map applied to the port or deletes all
the policy maps applied on the
ingress direction of the port . At
present, the egress does not support
the egress policy map.
pass-through-cos Forbid the packet to rewrite L2 CoS

1-10
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

no pass-through-cos value at the egress, the no command

allows the packet to rewrite L2 CoS
value.
pass-through-dscp Forbid the packet to rewrite dscp
no pass-through-dscp value at the egress, the no command
allows the packet to rewrite dscp
value.
Global Mode
Apply a policy map to the specified
VLAN interface; the no command
service-policy input <policy-map-name>
deletes the specified policy map
vlan <vlan-list>
applied to the VLAN interface or
no service-policy input
deletes all the policy maps applied in
{<policy-map-name>} vlan <vlan-list>
the ingress direction of the vlan
interface .

4. Configure queue management algorithm and weight

Command Explanation
Port Configuration Mode
mls qos queue algorithm {sp | wdrr} Set queue management algorithm, the
no mls qos queue algorithm default queue management algorithm
is wdrr.
Global Mode
mls qos queue wdrr weight Set wdrr queue weight for all ports
<weight0..weight7> globally, the default queue weight is 1
no mls qos queue wdrr weight 1 1 1 1 1 1 1.

5. Configure QoS mapping

Command Explanation
Global Mode
mls qos map {cos-intp <intp1…intp8> | Set the priority mapping for QoS, the
dscp-intp <in-dscp list> to <intp> | intp-cos no command restores the default
<intp list> to <out-cos> | intp-dp <intp list> mapping value.
to <out-dp> | intp-dscp <intp list> to
<out-dscp>| intp-intp <color> <intp list> to
<intp> | intp-queue <intp list> to
<out-queue> }
no mls qos map {cos-intp | dscp-intp |

1-11
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

intp-cos | intp-dp | intp-dscp | intp-intp

<color> | intp-queue}

6. Clear accounting data of the specific ports or VLANs

Command Explanation
Admin Mode
clear mls qos statistics [interface Clear accounting data of the
<interface-name> | vlan <vlan-id>] specified ports or VLAN Policy Map.
If there are no parameters, clear
accounting data of all policy map.

7. Show configuration of QoS

Command Explanation
Admin Mode
show mls qos maps [cos-intp | dscp-intp | Display the configuration of QoS
intp-intp | intp-cos | intp-dscp | intp-dp | mapping.
intp-queue]
show class-map [<class-map-name>] Display the classified map
information of QoS.
show policy-map [<policy-map-name>] Display the policy map information
of QoS.
show mls qos aggregate-policy Display the aggregate policy
[<aggregate-policy-name>] configuration of QoS.
show mls qos interface [<interface-id>] Display QoS configuration
[policy | queuing] information on a port.
show mls qos vlan <v-id> Display QoS configuration on VLAN
interface.

1.3 QoS Example

Example 1:
Enable QoS function, change the global queue out weight to 1:1:2:2:4:4:8:8, set
port ethernet 1/0/1 in trust CoS mode without changing DSCP value, and set the default
CoS value of the port to 5.
The configuration steps are listed below:
Switch#config

1-12
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

Switch(config)# mls qos queue weight 1 1 2 2 4 4 8 8

Switch(config)#interface ethernet 1/0/1
Switch(Config-If-Ethernet 1/0/1)#mls qos trust cos
Switch(Config-If-Ethernet 1/0/1)#pass-through-dscp
Switch(Config-If-Ethernet1/0/1)#mls qos cos 5

Configuration result:
When QoS enabled in Global Mode, the egress queue bandwidth proportion of each
port is 1:1:2:2:4:4:8:8. When packets have CoS value coming in through port
ethernet1/0/1, it will be map to the internal priority according to the CoS value, CoS value
0 to 7 correspond to queue out 1, 2, 3, 4, 5, 6, 7, 8 respectively. If the incoming packet has
no CoS value, it is default to 5 and will be put in queue6. All passing packets would not
have their DSCP values changed

Example 2:
In port ethernet1/0/2, set the bandwidth for packets from segment 192.168.1.0 to 10 Mb/s,
with a burst value of 4 MB, all packets exceed this bandwidth setting will be dropped.

The configuration steps are listed below:

Switch#config
Switch(config)#access-list 1 permit 192.168.1.0 0.0.0.255
Switch(config)#class-map c1
Switch(Config-ClassMap-c1)#match access-group 1
Switch(Config-ClassMap-c1)#exit
Switch(config)#policy-map p1
Switch(Config-PolicyMap-p1)#class c1
Switch(Config-PolicyMap-p1-Class-c1)#policy 10000 4000 exceed-action drop
Switch(Config-PolicyMap-p1-Class-c1)#exit
Switch(Config-PolicyMap-p1)#exit
Switch(config)#interface ethernet 1/0/2
Switch(Config-If-Ethernet1/0/2)#service-policy input p1

Configuration result:
An ACL name 1 is set to matching segment 192.168.1.0. Enable QoS globally, create
a class map named c1, matching ACL1 in class map; create another policy map named
p1 and refer to c1 in p1, set appropriate policies to limit bandwidth and burst value. Apply
this policy map on port ethernet1/0/2. After the above settings done, bandwidth for
packets from segment 192.168.1.0 through port ethernet 1/0/2 is set to 10 Mb/s, with a

1-13
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

burst value of 4 MB, all packets exceed this bandwidth setting in that segment will be
dropped.

Example 3:

Server
QoS area

Switch3

Switch2

Trunk

Switch1

Fig 1-7 Typical QoS topology

As shown in the figure, inside the block is a QoS domain, Switch1 classifies different
traffics and assigns different IP precedences. For example, set CoS precedence for
packets from segment 192.168.1.0 to 5 on port ethernet1/0/1(set the internal priority to 40,
set the default intp-dscp mapping to 40-40, the corresponding IP precedence to 5). The
port connecting to switch2 is a trunk port. In Switch2, set port ethernet 1/0/1 that
connecting to swtich1 to trust dscp. Thus inside the QoS domain, packets of different
priorities will go to different queues and get different bandwidth.

The configuration steps are listed below:

QoS configuration in Switch1:
Switch#config
Switch(config)#access-list 1 permit 192.168.1.0 0.0.0.255
Switch(config)#class-map c1
Switch(Config-ClassMap-c1)#match access-group 1
Switch(Config-ClassMap-c1)#exit
Switch(config)#policy-map p1
Switch(Config-PolicyMap-p1)#class c1
Switch(Config-PolicyMap-p1-Class-c1)#set ip precedence 40
Switch(Config-PolicyMap-p1-Class-c1)#exit

1-14
QoS and Flow-based
Redirection Configuration Chapter 1 QoS Configuration

Switch(Config-PolicyMap-p1)#exit
Switch(config)#interface ethernet 1/0/1
Switch(Config-If-Ethernet1/0/1)#service-policy input p1

QoS configuration in Switch2:

Switch#config
Switch(config)#interface ethernet 1/0/1
Switch(Config-If-Ethernet1/0/1)#mls qos trust dscp

1.4 QoS Troubleshooting

 trust cos and EXP can be used with other trust or Policy Map.
 trust dscp can be used with other trust or Policy Map. This configuration takes
effect to IPv4 and IPv6 packets.
 trust exp, trust dscp and trust cos may be configured at the same time, the
priority is: EXP>DSCP>COS.
 If the dynamic VLAN (mac vlan/voice vlan/ip subnet vlan/protocol vlan) is
configured, then the packet COS value equals COS value of the dynamic VLAN.
 At present, it is not recommended to synchronously use policy map on VLAN and
VLAN’s port.

1-15
QoS and Flow-based
Redirection Configuration Chapter 2 Flow-based Redirection

Chapter 2 Flow-based Redirection

2.1 Introduction to Flow-based Redirection

Flow-based redirection function enables the switch to transmit the data frames
meeting some special condition (specified by ACL) to another specified port. The fames
meeting a same special condition are called a class of flow, the ingress port of the data
frame is called the source port of redirection, and the specified egress port is called the
destination port of redirection. Usually there are two kinds of application of flow-based
redirection: 1. connecting a protocol analyzer (for example, Sniffer) or a RMON monitor
to the destination port of redirection, to monitor and manage the network, and diagnose
the problems in the network; 2. Special transmission policy for a special type of data
frames.
The switch can only designate a single destination port of redirection for a same
class of flow within a source port of redirection, while it can designate different
destination ports of redirection for different classes of flows within a source port of
redirection. The same class of flow can be applied to different source ports.

2.2 Flow-based Redirection Configuration Task

Sequence

1． Flow-based redirection configuration

2． Check the current flow-based redirection configuration
3． Configure the vlan redirection function of the port
4． Configure the maximum number of vlan of redirect on each port

1. Flow-based redirection configuration

Command Explanation
Physical Interface Configuration Mode
Specify flow-based
redirection for the port; the
access-group <aclname> redirect to interface
“no access-group
[ethernet <IFNAME>|<IFNAME>]
<aclname> redirect”
no access-group <aclname> redirect
command is used to delete
flow-based redirection.

2-1
QoS and Flow-based
Redirection Configuration Chapter 2 Flow-based Redirection

2. Check the current flow-based redirection configuration

Command Explanation
Global Mode/Admin Mode
Display the information of
show flow-based-redirect {interface [ethernet current flow-based
<IFNAME> |<IFNAME>]} redirection in the
system/port.

3. Configure the vlan redirection function of the port

Command Explanation
Port Mode
match vlan <1-4096> redirect interface (ethernet|)
IFNAME Configure the vlan redirection
function of the port.
no match vlan <1-4096> redirect
port-redirect match vlan <1-4094> source-port
interface (ethernet|) IFNAME destination-port
interface (ethernet|) IFNAME Configure the vlan redirection
no port-redirect match vlan <1-4094> source-port function of the port.
interface (ethernet|) IFNAME destination-port
interface (ethernet|) IFNAME

4. Configure the maximum number of vlan of redirect on each port

Command Explanation
Global Mode
Configure the maximum
vlan-port-redirect vlan maximum <1-1000>
number of vlan of redirect on
no vlan-port-redirect vlan maximum
each port.

2.3 Flow-based Redirection Examples

Example:
User’s request of configuration is listed as follows: redirecting the frames whose
source IP is 192.168.1.111 received from port 1 to port 6, that is sending the frames
whose source IP is 192.168.1.111 received from port 1 through port6.

Modification of configuration:

2-2
QoS and Flow-based
Redirection Configuration Chapter 2 Flow-based Redirection

1: Set an ACL, the condition to be matched is: source IP is 192.168.1.111;

2: Apply the redirection based on this flow to port 1.

The following is the configuration procedure:

Switch(config)#access-list 1 permit host 192.168.1.111
Switch(config)#interface ethernet 1/0/1
Switch(Config-If-Ethernet1/0/1)# access-group 1 redirect to interface ethernet 1/0/6

2.4 Flow-based Redirection Troubleshooting Help

When the configuration of flow-based redirection fails, please check that whether it
is the following reasons causing the problem:
 The type of flow (ACL) can only be digital standard IP ACL, digital extensive IP ACL,
nomenclature standard IP ACL, nomenclature extensive IP ACL, digital standard
IPv6 ACL, and nomenclature standard IPv6 ACL;
 Parameters of Timerange and Portrange can not be set in ACL, the type of ACL
should be Permit.
 The redirection port must be 1000Mb port in the flow-based redirection function.
 Do not implement the forward across VLAN for flow-based redirection.

2-3
QoS and Flow-based
Redirection Configuration Chapter 3 Egress QoS Configuration

Chapter 3 Egress QoS Configuration

3.1 Introduction to Egress QoS

In traditional IP networks, all packets are treated in the same way. All network
equipments treat them by the first-in-first-out policy and try best effort to send them to the
destination. However, it does not guarantee the performance like reliability and
transmission delay. Network develops so fast that new demand has been raised for the
quality of service on IP network with the continual emergence of new applications. For
example, delay-sensitive services like VoIP and video put higher demands on packet
transmission delay and users cannot accept too long transmission delay (by contrast,
E-mail and FTP services are not sensitive to the time delay). In order to support services
with different service requirement like voice, video and data service, the network is
required to be able to distinguish between different communications and provide
appropriate service. The traditional best-effort IP network cannot identify and distinguish
various kinds of communications while this ability is the very premise of providing
differentiated services for different communications. Therefore, the best-effort service
mode of traditional network cannot meet the demand of applications. The emergence of
QoS techniques is committed to solve this problem.
Egress PolicyMap is the QoS policy in egress which performs QoS control of packets
in the egress direction and provides better service for specified network communication
with kinds of techniques. Egress PolicyMap includes class-map and policy-map, of which
class-map is used for selecting packets to operate and policy-map is used for specifying
the operation to use. Not all equipments support Egress QoS currently.

3.1.1 Egress QOS Terms

Egress QoS: Achieving QoS on egress of port.
Inner_vid: VLAN ID brought by the TAG near the header of network layer when double
TAGs exist.
Outer_vid: VLAN ID brought by the TAG near the header of network link layer when
double TAGs exist. The TAG is considered to be outer tag by default when only one TAG
exists.
Outer_tpid: Protocol type of the network link layer header indicating the type of outer tag.

3.1.2 Basic Egress QoS Model

3-1
QoS and Flow-based
Redirection Configuration Chapter 3 Egress QoS Configuration

Ingress Egress

Generate
internal
priority color Policing and
Classification Policing Remark scheduling remark of
Egress

Sort packet traffic

according to the Decide whether traffic Place packets into priority Set the color of packet
classification info and color is single bucket Degrade or discard queues according to traffic according to policing
dual color or dual different color packets, internal priority, provide policy of Egress
convert classification info and remark DSCP, the service for them PolicyMap, degrade or
to internal priority value bucket three color
TOS, COS fields according to queue weight drop different color
and drop precedence according policing
and drop precedence packets
value policy

According to the characters (including field values like COS and DSCP) of upstream
packets, policing and rewriting of Egress make the last QoS change on the packet prior to
the packet egress.
Policing configures different policing policy based on the flow and distributes
bandwidth for the flow classified. The distribution policy of bandwidth can be either dual
bucket dual color or dual bucket three color. Different colors can be assigned to different
flows and approaches of discard or passage can be chosen for them; you can add
rewriting action for packets with passage approach chosen. See the following flow chart
for detailed description of Egress QoS:

3-2
QoS and Flow-based
Redirection Configuration Chapter 3 Egress QoS Configuration
Start

Classify or not No

Yes

Modify QoS
（optional）： Drop
Set cos/dscp

Transmit

Whether
configured No
Policing policy

Yes

Modify QoS according remark

table（optional）：
Cos-cos、cos-dscp、
dscp-cos、dscp-dscp
Action of red packets：
drop/transmit

结束

Description of action that modify QoS attribute according to egress remark table:
cos-cos：for cos value of packets, modify cos value of packets according to cos table of
QoS remarking
cos-dscp：for cos value of packets, modify dscp value of packets according to cos table of
QoS remarking
dscp-cos：for dscp value of packets, modify cos value of packets according to dscp table
of QoS remarking
dscp-dscp：for dscp value of packets, modify dscp value of packets according to dscp
table of QoS remarking

3.2 Egress QoS Configuration

Egress QoS Configuration Task List:

Configure class map
Set up a classification rule according to ACL, CoS, VLAN ID, IPv4 Precedent, DSCP,
IPV6 DSCP to classify the data stream. Different classes of data streams will be

3-3
QoS and Flow-based
Redirection Configuration Chapter 3 Egress QoS Configuration

processed with different policies.

Configure policy map
After data steam classification, a policy map can be created to associate with a class
map created earlier and enter policy class mode. Then different policies (such as
bandwidth limit, assigning new DSCP value) can be applied to different data streams.
Apply Egress QoS to port or VLAN
Configure the trust mode or binding policies for ports. A policy will only take effect on
a port when it is bound to that port.
The policy may be bound to the specific VLAN.
Set Egress QoS remark mapping
If modify QoS attribute by using Egress QoS remark in policy, it should set the
corresponding mapping. If it needs to take effect to green packets, modifying switch
of green packets should be enabled and ingress needs to trust the corresponding
QoS attribute (qos/dscp/exp).

1. Configure a class-map
Command Explanation
Global Mode
class-map <class-map-name> Create a class-map and enter
no class-map <class-map-name> class-map mode, no command
deletes the specified class-map.
match {access-group <acl-index-or-name> Configure the matched standard of the
| ip dscp <dscp-list> | ip precedence class map to classify the data stream
<ip-precedence-list> | ipv6 dscp according to ACL, CoS, VLAN ID,
<dscp-list> | vlan <vlan-list> | cos IPv4 Precedence, DSCP, IPv6 DSCP
<cos-list> | ipv6 access-group priority; no command deletes the
<acl-index-or-name>} specific matched standard.
no match {access-group | ip dscp | ip
precedence | ipv6 dscp | vlan | cos | ipv6
access-group}

2. Configure a policy-map
Command Explanation
Global Mode
Create a policy-map and enter
policy-map <policy-map-name>
policy-map mode, no command deletes
no policy-map <policy-map-name>
the specific policy-map.

3-4
QoS and Flow-based
Redirection Configuration Chapter 3 Egress QoS Configuration

class <class-map-name> Create a policy map to associate with a

[insert-before <class-map-name>] class map and enter policy class map
no class <class-map-name> mode, then different data streams can
apply different policies and be assigned a
new DSCP value. No command deletes
the specified policy class map.
set {ip dscp <new-dscp> | ip
precedence <new-precedence> | cos
Assign a new DSCP, CoS and IP
<new-cos> | c-vid <new-c-vid> | s-vid
Precedence value for the classified flow,
<new-s-vid> | s-tpid <new-s-tpid>}
no command cancels the operation.
no set {ip dscp | ip precedence | cos |
c-vid | s-vid | s-tpid}
Single bucket mode: Configure a policy for the classified flow.
policy <bits_per_second> The non-aggregation policy command
<normal_burst_bytes> ({action supports three colors. Analyze the
ACTION} | exceed-action drop | working mode of the token bucket,
transmit}) whether it is single rate single bucket,
single rate dual bucket or dual rate dual
Dual bucket mode: bucket, set corresponding action to
policy <bits_per_second> different color packets. The no command
<normal_burst_bytes> [pir will delete the configuration. Only specific
<peak_rate_bps>] | switch supports single bucket mode.
<maximum_burst_bytes> [{action
ACTION | violate-action drop |
transmit}]

ACTION definition:
policied-cos-to-cos-transmit |
policied-cos-to-dscp-transmit |
policied-dscp-exp-to-cos-transmit |
policied-dscp-exp-to-dscp-transmit
no policy
Set statistic function for the classified
flow. After enable this function under the
accounting policy class map mode, add statistic
no accounting function to the flow of the policy class
map. In single bucket mode, packets can
only red or green when passing policy. In

3-5
QoS and Flow-based
Redirection Configuration Chapter 3 Egress QoS Configuration

the print information, in-profile means

green and out-profile means red. In dual
bucket mode, there are three colors of
packets in-profile means green and
out-profile means red and yellow.

3. Apply policy to port or VLAN

Command Explanation
Interface Mode
service-policy output Apply a policy map to the egress of the
<policy-map-name> port; the no command deletes the
no service-policy output specified policy map applied to the port
{<policy-map-name>} or deletes all the policy maps applied on
the egress direction of the port .
Global Mode
service-policy output Apply a policy map to the egress of the
<policy-map-name> vlan <vlan-list> VLAN; the no command deletes the
no service-policy output specified policy map applied to the VLAN
{<policy-map-name>} vlan <vlan-list> interfaceor deletes all the policy maps
applied in the egress direction of the vlan
interface .

4． Set Egress QoS remark mapping

Command Explanation
Global Mode
mls qos map {cos-cos | cos-dscp} {green | Set Egress cos mapping, no
yellow | red} <value1> <value2>…<value8> command resotores the default
no mls qos map {cos-cos | cos-dscp} configuration.
{green | yellow | red}
mls qos map {dscp-cos | dscp-dscp} Set Egress dscp mapping, <dscp-list>
{green | yellow | red} <dscp list> to <value> means 1 to 8 dscp values, no
no mls qos map {dscp-cos | dscp-dscp} command restores the default
{green | yellow | red} configuration.
mls qos egress green remark Set Egress QoS remark mapping to
no mls qos egress green remark take effect for green packets, no
command does not take effect to
green packets.

3-6
QoS and Flow-based
Redirection Configuration Chapter 3 Egress QoS Configuration

5. Clear accounting data of the specific ports or VLANs

Command Explanation
Admin Mode
clear mls qos statistics [interface Clear accounting data of the
<interface-name> | vlan <vlan-id>] specified ports or VLAN Policy Map.
If there are no parameters, clear
accounting data of all policy map.

6. Show QoS configuration

Command Explanation
Admin Mode
show mls qos {interface [<interface-id>] Show QoS configuration of the port.
[policy | queuing] | vlan <vlan-id>}

show class-map [<class-map-name>] Show the class map information of QoS.

Show the policy map information of

show policy-map [<policy-map-name>]
QoS.
show mls qos maps {cos-cos | cos-dscp |
Show mapping relation of Egress QoS
dscp-cos | dscp-exp} {green | yellow | red
remark.
|}

3.3 Egress QoS Examples

Example1:
On the egress of the port1, change cos value as 4 for the packet with dscp value of 0.
Create a class map:
switch(config)#class-map 1
switch(config-classmap-1)#match ip dscp 0
switch(config-classmap-1)#exit

Create a policy map:

switch(config)#policy-map 1
switch(config-policymap-1)#class 1
switch(config-policymap-1-class-1)#set cos 4
switch(config-policymap-1-class-1)#exit
switch(config-policymap-1)#exit

3-7
QoS and Flow-based
Redirection Configuration Chapter 3 Egress QoS Configuration

Bind a policy to the port:

switch(config)#in e 1/0/1
switch(config-if-ethernet1/0/1)#service-policy output 1

Example2:
On the egress of vlan10, change cos value as 4 for the packet with ipv6 dscp value of
7.
Create a class map:
switch(config)#class-map 1
switch(config-classmap-1)#match ipv6 dscp 7
switch(config-classmap-1)#exit

Create a policy map:

switch(config)#policy-map 1
switch(config-policymap-1)#class 1
switch(config-policymap-1-class-1)#set cos 4
switch(config-policymap-1-class-1)#exit
switch(config-policymap-1)#exit

Bind a policy to VLAN

switch(config)#service-policy output 1 vlan 10

Example 3:
In egress of port 1, limit the speed of packets. Set the bandwidth for packets to 1 Mb/s,
with the normal burst value of 1 MB, the max burst value of 4 MB, set dscp value of 1 as
10 for green packets, set dscp value of yellow packets as 9 and drop red packets.
Create a class map
switch(config)#class-map c1
switch(config-classmap-c1)#match ip dscp 1
switch(config-classmap-c1)#exit
Create a policy map
switch(config)#policy-map p1
switch(config-policymap-p1)#class c1
switch(config-policymap-p1-class-c1)#policy 1000 1000 4000 action
policied-dscp-exp-to-dscp-transmit violate-action drop
switch(config-policymap-p1-class-c1)#exit
switch(config-policymap-p1)#exit

3-8
QoS and Flow-based
Redirection Configuration Chapter 3 Egress QoS Configuration

Set Egress dscp remark mapping

switch(config)#mls qos map dscp-dscp green 1 to 10
switch(config)#mls qos map dscp-dscp yellow 1 to 9
Set Egress remark to take effect for green packets
switch(config)#mls qos egress green remark
Set trust dscp mode on ingress
switch(config-if-port-range)#mls qos trust dscp
Bind policy to egress of port1
switch(config-if-ethernet1/0/1)#service-policy output p1

3.4 Egress QoS Troubleshooting Help

 Not all equipments support Egress QoS presently, so please make sure the current
device supports this function.
 If the policy configured cannot bind to the port or VLAN, please check whether the
match option in classification table is supported by the current device.
 If terminal printing suggests lack of resource, please make sure there is enough
resource to send the current policy.
 If the policy with match acl configured cannot bind to the port or VLAN, please make
sure rules including permit exist in ACL.
 If modifying QoS attribute is invalid by Egress QoS remark, please ensure whether
ingress sets the correspondsing QoS attribute with trust.
 If egress set QoS attributes (set cos/ip dscp) for modifying all packets, and it uses
Egress remark to modify QoS attributes for packets of different colors, previous
modification is preferential for modifying packets.

3-9
QoS and Flow-based
Redirection Configuration Chapter 4 Flexible QinQ Configuration

Chapter 4 Flexible QinQ Configuration

4.1 Introduction to Flexible QinQ

4.1.1 QinQ Technique

Dot1q-tunnel is also called QinQ (802.1Q-in-802.1Q), which is an expansion of
802.1Q. Its dominating idea is encapsulating the customer VLAN tag (CVLAN tag) to the
service provider VLAN tag (SPVLAN tag). The packet with two VLAN tags is transmitted
through the backbone network of the ISP internet to provide a simple layer-2 tunnel for the
users. It is simple and easy to manage, applicable only by static configuration, and
especially adaptive to small office network or small metropolitan area network using
layer-3 switch as backbone equipment.
There are two kinds of QinQ: basic QinQ and flexible QinQ, the priority of flexible
QinQ is higher than basic QinQ.

4.1.2 Basic QinQ

Basic QinQ based the port. After a port configures QinQ, whether the received packet
with tag or not, the device still packs the default VLAN tag for the packet. Using basic
QinQ is simple, but the setting method of VLAN tag is inflexible.

4.1.3 Flexible QinQ

Flexible QinQ based data flow. It selects whether pack the external tag and packs
what kind of the external tag by matching the material flow. For example: implement the
property of flexible QinQ according to the user’s VLAN tag, MAC address, IPv4/IPv6
address, IPv4/IPv6 protocol and the port ID of the application, etc. So, it can encapsulate
the external tag for the packet and implements different scheme by different users or
methods.

4.2 Flexible QinQ Configuration Task List

The match of flexible QinQ data flow uses policy-map rule of QoS to be sent, the
configuration task list is as follows:
1. Create class-map to classify different data flows

4-1
QoS and Flow-based
Redirection Configuration Chapter 4 Flexible QinQ Configuration

2. Create flexible QinQ policy-map to relate with the class-map and set the corresponding
operation
3. Bind flexible QinQ policy-map to port

1. Configure class map

Command Explanation
Global mode
class-map <class-map-name> Create a class-map and enter
no class-map <class-map-name> class-map mode, the no command
deletes the specified class-map.
match {access-group <acl-index-or-name> Set the match standard of class-map,
| ip dscp <dscp-list>| ip precedence (classify data flow by ACL, CoS, VLAN
<ip-precedence-list>| ipv6 access-group ID, IPv4 Precedent or DSCP, etc for
<acl-index-or-name>| ipv6 dscp the class map); the no command
<dscp-list> | ipv6 flowlabel <flowlabel-list> deletes the specified match standard.
| vlan <vlan-list> | cos <cos-list>}
no match {access-group | ip dscp | ip
precedence|ipv6 access-group| ipv6 dscp |
ipv6 flowlabel | vlan | cos}

2. Configure policy-map of flexible QinQ

Command Explanation
Global mode
policy-map <policy-map-name> Create a policy-map and enter
no policy-map <policy-map-name> policy-map mode, the no command
deletes the specified policy-map.
class <class-map-name> [insert-before After a policy-map is created, it can
<class-map-name>] be associated to a class. Different
no class <class-map-name> policy or new DSCP value can be
applied to different data flows in class
mode; the no command deletes the
specified class-map.
set s-vid < vid> Set external VLAN Tag for the
no set s-vid classified traffic, no command
cancels the operation.
add s-vid <vid> Add external VLAN Tag for the
no add s-vid <vid> classified traffic, no command
cancels the operation.

4-2
QoS and Flow-based
Redirection Configuration Chapter 4 Flexible QinQ Configuration

3. Bind flexible QinQ policy-map to port

Command Explanation
Port mode
service-policy input<policy-map-name> Apply a policy-map to a port, the no
no service-policy input<policy-map-name> command deletes the specified
policy-map applied to the port.
Global mode

service-policy input<policy-map-name> Apply a policy-map to a VLAN, the no

vlan<vid> command deletes the specified
no service-policy input<policy-map-name> policy-map applied to the VLAN.
vlan <vid>

4. Show flexible QinQ policy-map bound to port

Command Explanation
Admin mode
show mls qos {interface [<interface-id>] Show flexible QinQ configuration on the
port.

4.3 Flexible QinQ Example

Fig 4-1 Flexible QinQ application topology

4-3
QoS and Flow-based
Redirection Configuration Chapter 4 Flexible QinQ Configuration

As shown in the figure, the first user is assigned three VLANs that the tag values are
1001, 2001, 3001 respectively in DSLAM1. VLAN1001 corresponds to Broad Band
Network, VLAN2001 corresponds to VOIP, VLAN3001 corresponds to VOD. After the
downlink port enables flexible QinQ function, the packets will be packed with different
external tags according to VLAN ID of users. The packet with tag 1001 will be packed an
external tag 1001 directly(This tag is unique in public network), enter Broad Band
Network-VLAN1001 and classfied to BRAS device. The packet with tag 2001(or 3001) will
be packed an external tag 2001(or 3001) and classfied to SR device according to the flow
rules. The second user can be assigned different VLAN tags for different VLANs in
DSLAM2. Notice: The assigned VLAN tag of the second user may be same with the first
user and the packet with tag will be also packed an external tag. In the above figure, the
external tag of the second user is different to the first user for distinguishing DSLAM
location and locating the user finally.
The configuration in the following:
If the data flow of DSLAM1 enters the switch’s downlink port1, the configuration is as
follows:
Switch(config)#class-map c1
Switch(config-classmap-c1)#match vlan 1001
Switch(config-classmap-c1)#exit
Switch(config)#class-map c2
Switch(config-classmap-c2)#match vlan 2001
Switch(config-classmap-c2)#exit
Switch(config)#class-map c3
Switch(config-classmap-c3)#match vlan 3001
Switch(config-classmap-c3)#exit
Switch(config)#policy-map p1
Switch(config-policymap-p1)#class c1
Switch(config-policymap-p1-class-c1)# set s-vid 1001
Switch(config-policymap-p1)#class c2
Switch(config-policymap-p1-class-c2)# set s-vid 2001
Switch(config-policymap-p1)#class c3
Switch(config-policymap-p1-class-c3)# set s-vid 3001
Switch(config-policymap-p1-class-c3)#exit
Switch(config-policymap-p1)#exit
Switch(config)#interface ethernet 1/0/1
Switch(config-if-ethernet1/0/1)#service-policy input p1
If the data flow of DSLAM2 enters the switch’s downlink port1, the configuration is as
follows:

4-4
QoS and Flow-based
Redirection Configuration Chapter 4 Flexible QinQ Configuration

Switch(config)#class-map c1
Switch(config-classmap-c1)#match vlan 1001
Switch(config-classmap-c1)#exit
Switch(config)#class-map c2
Switch(config-classmap-c2)#match vlan 2001
Switch(config-classmap-c2)#exit
Switch(config)#class-map c3
Switch(config-classmap-c3)#match vlan 3001
Switch(config-classmap-c3)#exit
Switch(config)#policy-map p1
Switch(config-policymap-p1)#class c1
Switch(config-policymap-p1-class-c1)# set s-vid 1002
Switch(config-policymap-p1)#class c2
Switch(config-policymap-p1-class-c2)# set s-vid 2002
Switch(config-policymap-p1)#class c3
Switch(config-policymap-p1-class-c3)# set s-vid 3002
Switch(config-policymap-p1-class-c3)#exit
Switch(config-policymap-p1)#exit
Switch(config)#interface ethernet 1/0/1
Switch(config-if-ethernet1/0/1)# service-policy input p1

4.4 Flexible QinQ Troubleshooting

If flexible QinQ policy can not be bound to the port, please check whether the problem
is caused by the following reasons:
 Make sure flexible QinQ whether supports the configured class-map and policy-map
 Make sure ACL includes permit rule if the class-map matches ACL rule
 Make sure the switch exists enough TCAM resource to send the binding
 Priority of flexible QinQ and vlan ingress filtering for processing packets is: flexible
QinQ > vlan ingress filtering

4-5