238 Chapter 7: QoS Features Available on the Catalyst 4000 IOS Family of Switches and the Catalyst
queue as a high-priority queue forces the switch to service packets in transmit queue 3 first,
and thus reduces the chance of latency or jitter for VoIP traffic. All interfaces that connect
to other switches and Cisco IP Phone ports use transmit queue 3 as a high-priority queue.
Finally, the configuration consists of traffic shaping of transmit queues connected to an
edge router. The edge router only routes data traffic and therefore traffic-shaping works well
in this example. The edge router connects to an ATM cloud, which can service only about
3.0 Mbps of data. As a result, the traffic-shaping configuration limits traffic to specific rates
per queue.
QoS Support on the Catalyst 2948G-L3, 4908G-L3, and
Catalyst 4000 Layer 3 Services Module
The following sections discuss the Catalyst 2948G-L3 and 4908G-L3 switches, and the
Catalyst 4000 Layer 3 services module. These switches resemble IOS-based routers from
a configuration perspective and are generally referred to as G-L3 switches. However, these
switches support only a few QoS features. The supported QoS features include rate
limiting, IOS-based traffic shaping, and output scheduling. These switches base classifi-
cation just on IP precedence values and have no support for marking.
These features use hardware TCAM to achieve 6 million packets-per-second performance.
As with other Catalyst platforms, the term hardware switching represents using high-speed
hardware components such as TCAM for packet processing.
These Layer 3 switches provide for classification based on IP precedence only. These
switches do not support classification based on DSCP or classification based on the Layer
2 CoS field of a packet. Furthermore, these Layer 3 switches do not support classification
based on ACLs, reclassification, or marking. As a result, these Layer 3 switches function
solely as a QoS packet-forwarding switch. In other words, the intended QoS function of
these Layer 3 switches is to route and schedule packets. In an end-to-end design, these
switches rely on the adjoining switches and edge routers for classification and marking.
This part this chapter covers the following topics and QoS features for the G-L3 switches:
• Architecture Overview
• Software Requirements
• Global Configuration
• Classification
• Output Scheduling
• Per-Port Traffic Shaping
• Rate Limiting
• Case Study
QoS Support on the Catalyst 2948G-L3, 4908G-L3, and Catalyst 4000 Layer 3 Services Module 239
Catalyst 2948G-L3, 4908G-L3, and 4232-L3 Services Module QoS
Architectural Overview
The 2948G-L3 and 4908G-L3 Layer 3 switches and the WS-X4232-L3 Layer 3 services
module share the same architecture with different port densities. The architecture of these
switches uses TCAM memory for packet processing and forwarding of IP, IPX, and Layer
2 frames. As with the Catalyst 4000 IOS Family of switches, packet processing and
forwarding needs to occur via hardware switching rather than software switching.
Hardware switching allows for high-rate traffic flows with several QoS features. The QoS
features supported on these switches include weighted round-robin (WRR) scheduling,
traffic policing, and traffic shaping for IP packets. QoS features do not exist for IPX and
bridged traffic, and these QoS features cannot coexist with IPX configurations.
The WS-X4232-L3 Services Module Architecture
The Catalyst 4000 Layer 3 services module, WS-X4232-L3, employs a unique architecture
for integration into the Catalyst 4000 CatOS switch. At the time of publication, the Catalyst
4000 IOS Family of switches does not support the Catalyst 4000 Layer 3 services module.
The WS-X4232-L3 module consists of 4 Gigabit Ethernet ports and 32 10/100-Mbps ports.
Two of the four Gigabit Ethernet ports connected directly to the supervisor interconnect on
the Catalyst 4000 or 4500 chassis backplane. The other two Gigabit Ethernet ports are
available as front-panel ports. The architecture of WS-X4232-L3 modules intends for the
two front-panel ports to function as Layer 3 ports only. All four Gibabit Ethernet ports
support port channeling (Gigabit EtherChannel). However, port-channel interfaces do not
support QoS features. Furthermore, the WS-X4232-L3 does not support integrated routing
and bridging (IRB). The 32 10/100-Mbps front panels are Layer 2-only ports. These 32 10/
100-Mbps front panels are configurable only from the CatOS command-line interface
(CLI) and not from the WS-X4232-L3 services module. Figure 7-7 provides a logical
depiction of the WS-X4232-L3 module architecture when configured in a Catalyst 4000
CatOS switch.
The configuration of the G-L3 switches is rather unique, with many restrictions because the
switch configures just like an IOS router. For instance, VLAN interfaces do not exist in the
configuration. Because of the unique configuration of these Layer 3 switches, consult the
following technical documents at Cisco.com for more details regarding configurations,
limitations, and caveats:
• “Configuration and Overview of the Router Module for the Catalyst 4000 Family
(WS-X4232-L3)” Document ID: 6198
• “Catalyst 4908G-L3 VLAN Routing and Bridging Example Configuration”
Document ID: 14972
• “Catalyst 2948G-L3 Sample Configurations - Single VLAN, Multi-VLAN, and
Multi-VLAN Distribution Layer Connecting to Network Core” Document ID: 12020
240 Chapter 7: QoS Features Available on the Catalyst 4000 IOS Family of Switches and the Catalyst
Figure 7-7 WS-X4232-L3 in a Catalyst 4000 Switch Architecture Depiction
Software Requirements
All software versions for the Catalyst 2948G-L3 and 4908G-L3 Layer 3 switches and the
WS-X4232-L3 Layer 3 services support output scheduling using WRR. Cisco IOS
Software versions 12.0(10)W5(18e) and later add feature support for per-port traffic
shaping and rate limiting.
Catalyst 4000 CatOS switches require Cisco CatOS Software version 5.5 or higher for use
of the WS-X4232-L3 module.
Global Configuration
The Catalyst 2948G-L3 and 4908G-L3 Layer 3 switches and the WS-X4232-L3 module
schedule packets based on IP precedence by default. Use the following command to disable
QoS on these switches:
[no] qos switching
The show qos switching command displays the state of global QoS configuration. The
show qos mapping outputs the global QoS WRR mapping configuration. The show qos
mapping [destination egress-interface] command displays the QoS WRR mapping
configuration for an egress interface.
<slot>/1 <slot>/2
G3 G4
G2G1
2 (G1 and G2) 1 Gbps
Layer 3 Ports
32 10/10 Mbps Ports
4,1 Gbps Backplane
Interconnects
Logical Diagram of
WS-X4232-L3
QoS Support on the Catalyst 2948G-L3, 4908G-L3, and Catalyst 4000 Layer 3 Services Module 241
Example 7-29 shows samples of the show qos switching and show qos mapping
commands, respectively.
Classification
The Catalyst 2948G-L3 and 4908G-L3 Layer 3 switches and the WS-X4232-L3 module
output schedule packets using a four-queue class model. Classification differentiates
packets into one of four queue classes for output scheduling.
The Catalyst 2948G-L3 and 4908G-L3 Layer 3 switches and the WS-X4232-L3 module
derive classification solely based on IP precedence information. Furthermore, these
switches classify frames based on the two most-significant bits (MSBs) of the IP prece-
dence field. Table 7-8 illustrates how IP precedence bits map to the queue classes.
Output Scheduling
The Catalyst 2948G-L3 and 4908G-L3 Layer 3 switches and the WS-X4232-L3 module
support output scheduling using WRR scheduling, rate limiting, and traffic-shaping
features. These switches support traffic shaping only on egress traffic flows. Rate-limiting
support is applicable to either ingress or egress traffic flows. The following sections discuss
WRR scheduling, rate limiting, and traffic shaping applicable to the Catalyst 2948G-L3 and
4908G-L3 Layer 3 switches and the WS-X4232-L3 module.
Example 7-29 Sample Use of show qos switching and show qos mapping Commands
Router#show qos switching
qos Based IP Switching enabled
Router#show qos mapping destination GigabitEthernet 1
Precedence WRR-Weight
0 1
1 2
2 3
3 4
Table 7-8 Default IP Precedence–to–Queue Class Mapping
IP Precedence (Binary) Queue Class
0 (000) Queue 0
1 (001) Queue 0
2 (010) Queue 1
3 (011) Queue 1
4 (100) Queue 2
5 (101) Queue 2
6 (110) Queue 3
7 (111) Queue 3
242 Chapter 7: QoS Features Available on the Catalyst 4000 IOS Family of Switches and the Catalyst
WRR Scheduling
WRR scheduling imparts higher bandwidth to higher-priority queues while still providing service
to lower-priority queues. Maintaining service to lower-priority queues avoids queue starvation.
The premise for WRR on the Catalyst 2948G-L3 and 4908G-L3 Layer 3 switches and the
WS-X4232-L3 module is to apply effective bandwidths to each of the four queues. Use the
following effective bandwidth formula to determine WRR weight:
(W/S)x B = n
W represents the WRR weight of the queue, and S represents the sum of all weights of the
active queues. B is the available bandwidth of the outgoing interface(s), and n represents
the effective bandwidth. Table 7-9 displays the default effective bandwidth per queue.
Chapter 2 discusses the behavior of WRR in more detail.
The queue mapping of the two MSBs of IP precedence to the WRR weighted value configures
globally and per interface. Interface configuration overrides the global configuration. Use the
following command to globally adjust the IP precedence to the WRR weighted value:
qos mapping precedence
value
wrr-weight
weight
To adjust the mapping of IP precedence to the WRR weighted value, use the following
global configuration command:
qos mapping [destination
egress-interface
] precedence
value
wrr-weight
weight
For both commands, value defines the two MSBs of IP precedence, and weight represents
the WRR weight. egress-interface defines the egress interface to apply the command
configuration.
Table 7-9 Default WRR Weights Assigned to Queue Class
Queue
Class
Two MSB of IP
Precedence Weight Effective Bandwidth
Percent of Total
Bandwidth
Queue 0 0 1 100 Mbps 10%
Queue 1 1 2 200 Mbps 20%
Queue 2 2 3 300 Mbps 30%
Queue 3 3 4 400 Mbps 40%
Example 7-30 Sample Configuration of Mapping IP Precedence to WRR Weights Globally and Per Interface
Router#show running-config
Building configuration...
(text deleted)
qos mapping precedence 1 wrr-weight 1
qos mapping destination GigabitEthernet1 precedence 2 wrr-weight 1
(text deleted)
!
end
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset