CBWFQ and LLQ
CBWFQ
Class-Based Weighted Fair Queuing (CBWFQ) addresses some of the problems
with WFQ. It allows manual configuration of traffic classification and
minimum bandwidth guarantees. It uses the MQC to configure classes and
allocate bandwidth. Therefore, you can group traffic into classes based on
any of the criteria available through the MQC. Each class becomes its own
FIFO queue at the interface. This is still considered Weighted Fair Queuing
because each queue is assigned a weight based on the class bandwidth guarantee,
and the scheduler takes packets from each queue based on those
weights.
There are three ways to designate bandwidth within an MQC policy map.
All the CBWFQ classes within a single policy map must use the same
method.
■ The bandwidth bandwidth-in-kbps command
■ The bandwidth percent command
■ The bandwidth remaining percent command
Each of these methods designates a minimum bandwidth to allocate to the
class. Traffic is allowed to burst above that amount if extra bandwidth is
available. You can allocate up to 75 percent of the interface bandwidth by
default; the rest is reserved for system traffic, such as routing updates. This
can be changed using the max-reserved-bandwidth interface command.
You can optionally specify the maximum number of packets allowed in each
class queue with the policy map command queue-limit number-of-packets.
The default is 64
Each policy map has a default class that is created automatically. All traffic
not identified by one of the other classes is placed in this queue. You can
allocate bandwidth to this class or enable WFQ for its traffic, but not both.
To enable WFQ, use the fair-queue command. If WFQ is enabled, you can
also configure the number of dynamic queues with the fair-queue numberof-
queues command.
Benefits of CBWFQ include:
■ Control over traffic classification
■ Minimum bandwidth guarantees
■ Granular control and scalability
Drawbacks of CBWFQ include:
■ Voice traffic can be delayed