Tail drop is a congestion avoidance technique that drops packets when an output queue is full, until congestion
is eliminated.
Use the
show qos interface
command to display the queue limit and other QoS values.
Queue Limit Default Values
These default values are used when
queue-limit
is not configured in the class:
• If QoS is configured and Weighted Random Early Detection (WRED) is not configured:
• Queue limit is 10 ms at the guaranteed service rate of the queue for non-priority queues.
• Queue limit is 10 ms at the interface rate for priority classes
• If QoS is configured and WRED is configured:
• Queue limit is two times the WRED maximum threshold. The maximum threshold can be an
explicitly configured value or an implicit 10 ms.
• If more than one WRED profile is configured in the class, the maximum threshold is the maximum
for all profiles.
• When the
queue-limit
is configured in time units, the guaranteed service rate (for the non-priority
class) or the interface rate (for the priority class) is used to compute the queue-limit.
These restrictions apply to queue limits:
• For releases before Release 6.3.2, the queue limit should be at least the maximum MTU size, which is
fixed at 9 * 1024 bytes = 9kb. From Release 6.3.2 onwards, the minimum queue limit is the interface
MTU (which is dynamically configured).
• Queue limit should be 3 GB, which is the maximum packet buffer size in ingress and egress queuing
ASICs.
• Only time-based units are allowed on bundle targets.
Guaranteed Service Rate
The guaranteed service rate is defined as the service rate of the queue when all queues are backlogged and
derived as:
minimum_ban (bandwidth_remaining_percent * unallocated_bandwidth)
This example shows the guaranteed service rate calculation:
policy-map sample_policy
class c1
bandwidth percent 30
bandwidth remaining percent 40
class c2
bandwidth percent 20
class class-default
guaranteed service rate of c1 = 30 percent LR + (40 percent * 50 percent * LR)
guaranteed service rate of c2 = 20 percent LR + (30 percent * 50 percent * LR)
guaranteed service rate of class-default = 30 percent * 50 percent * LR
Modular QoS Command Reference for Cisco NCS 5500 Series and Cisco NCS 540 Series Routers
69
Congestion Avoidance Commands
queue-limit