26-5
Catalyst 3750 Metro Switch Software Configuration Guide
78-15870-01
Chapter 26 Configuring QoS
Understanding QoS
• Marking evaluates the policer configuration information for the action to take when a packet is out
of profile. Marking actions are to pass through a packet without modification, to mark down the QoS
label in the packet, or to drop the packet. For more information, see the “Ingress Policing and
Marking” section on page 26-9.
• Queueing evaluates the QoS label and the corresponding DSCP or CoS value to select into which of
the two ingress queues to place a packet. Queueing is enhanced with the WTD algorithm, a
congestion-avoidance mechanism. If the threshold is exceeded, the packet is dropped. For more
information, see the “Queueing and Scheduling Overview” section on page 26-12.
• Scheduling services the queues based on their configured (SRR) weights. One of the ingress queues
is the priority queue, and SRR services it for its configured share before servicing the other queue.
For more information, see the “SRR Shaping and Sharing” section on page 26-14.
These are the actions when the traffic is sent out a standard port:
• Queueing evaluates the QoS label and the corresponding DSCP or CoS value to select into which
queue-set (a set of four queues per port) to place a packet. Because congestion can occur when
multiple ingress ports simultaneously send data to an egress port, WTD is used to differentiate traffic
classes and to subject the packets to different thresholds based on the QoS label. If the threshold is
exceeded, the packet is dropped. For more information, see the “Queueing and Scheduling
Overview” section on page 26-12.
• Scheduling services the four egress queues based on their configured SRR shared or shaped weights.
These are the actions when the traffic is sent out an ES port:
• Classification is the process of generating a distinct path for a packet by matching the CoS, DSCP,
IP precedence, MPLS EXP bits in the header, or matching a packet based on the inner and the outer
VLAN IDs. The hierarchical configuration controls the number of class-level, VLAN-level, and
physical-interface-level queues to create. For information, see the “Understanding Hierarchical
QoS” section on page 26-19 and the “Hierarchical Levels” section on page 26-20. For classification
information, see the “Egress Classification Based on Traffic Classes and Traffic Policies” section on
page 26-23.
• Policing decides whether a packet is in or out of profile by comparing the rate of the outbound traffic
to the configured policer. The policer limits the bandwidth consumed by a flow of traffic. The result
is passed to the marker. For more information, see the “Egress Policing and Marking” section on
page 26-24.
• Marking evaluates the policer and configuration information for the action to be taken when a packet
is out of profile and decides what to do with the packet (pass through a packet without modification,
mark down the QoS label in the packet, or drop the packet). For more information, see the “Egress
Policing and Marking” section on page 26-24.
• Queueing is accomplished through a hierarchical queueing framework, in which the switch assigns
each packet to a queue based on the packet class, VLAN, and physical interface. Tail drop or WRED
can be configured per queue as the congestion-avoidance mechanism. With tail drop, packets are
queued until the maximum threshold is exceeded, and then all the packets are dropped. WRED
reduces the chances of tail drop by selectively dropping packets when the port begins to show signs
of congestion. For more information, see the “Queueing and Scheduling of Hierarchical Queues”
section on page 26-26.
• Scheduling is accomplished through CBWFQ or LLQ (strict priority queueing). CBWFQ is a
mechanism that provides guaranteed bandwidth to a particular traffic class while still fairly serving
all other traffic in the network. LLQ is another scheduling mechanism, which ensures that
delay-sensitive traffic is queued and sent before the traffic in other queues. Scheduling services the
queues through average-rate shaping. For more information, see the “Queueing and Scheduling of
Hierarchical Queues” section on page 26-26.
To Next Page
Loading...
