QoS Support on the Catalyst 2948G-L3, 4908G-L3, and Catalyst 4000 Layer 3 Services Module 243
To demonstrate WRR behavior, several trials were performed using the topology in Figure 7-8.
In each trial, a traffic generator transmitted traffic to the switch at 1 Gbps on both supervisor
ports 1/1 and 1/2. The ingress traffic on port 1/1 had an IP precedence value of 0, and ingress
traffic on port 1/2 had an IP precedence value of 7. The traffic was routed from interfaces 1/1
and 1/2 out to interface Gigabit Ethernet 1 on the WS-X4232-L3 module. The traffic
generated was also connected to this port to measure the egress traffic rate for both streams.
In the first trial, the default QoS mapping configuration was used. In the second trial, the
QoS mapping configuration was changed such that traffic from both streams mapped to the
same WRR weight. Table 7-10 summarizes the results of the trial.
Based on the WRR bandwidth formula, both trials yielded expected results. For the default
WRR mapping trial, the router used only two queues. As a result, the sum of all weighted values
was 5 (because Q1 has a weight of 1 and Q4 has a weight of 4). Therefore, the derived formula
for the effective bandwidth for the IP precedence traffic of 7 was computed as follows:
(W/S)x B = n
(4/5) x 1.0 Gbps = 800 Mbps for IP precedence 7 traffic
For the second trial, mapping all the IP precedence values to the same WRR weight yields
an even distribution of egress traffic for both streams.
Per-Port Traffic Shaping
Traffic shaping uses the leaky token bucket algorithm. The leaky token bucket algorithm
discussed earlier in this chapter for the Catalyst 4000 IOS Family also applies to the
Catalyst 2948G-L3 and 4908G-L3 Layer 3 switches and the WS-X4232-L3. Although the
hardware architecture and leaky token bucket algorithm implementations differ signifi-
cantly between the two types of switches, the algorithm still applies for understanding the
rate and burst parameters for the Catalyst 2948G-L3 and 4908G-L3 Layer 3 switches and
the WS-X4232-L3 Layer 3 services switches. Note that unlike rate-limiting policers, traffic
shaping actually buffers packets that exceed the specified rate.
Use the following command to configure traffic shaping in the interface configuration mode:
traffic-shape
rate burst
Table 7-10 WRR on the Catalyst WS-X4232-L3 Module
Trial Default WRR Mapping
All IP Precedence Map
to the Same WRR Weight
Measured rate of traffic for frames
with IP precedence (0)
200 Mbps 500 Mbps
Measured rate of traffic for frames
with IP precedence (7)
800 Mbps 500 Mbps