Table Of Contents
QoS: Implicit Queue Limit
The QoS: Implicit Queue Limit feature eliminates the necessity to manually configure queue limits by automatically configuring queue limits if they are not explicitly configured. Previously, default configurations did not protect queues on Cisco 12000 Series ISE line cards against oversubscription and starvation.
Feature History for QoS: Implicit Queue Limit
This feature was introduced for the Cisco 12000 Series ISE line cards (Engine 3) with Gigabit Ethernet and packet-over-SONET interfaces.
Finding Support Information for Platforms and Cisco IOS Software Images
Use Cisco Feature Navigator to find information about platform support and Cisco IOS software image support. Access Cisco Feature Navigator at http://www.cisco.com/go/fn. You must have an account on Cisco.com. If you do not have an account or have forgotten your username or password, click Cancel at the login dialog box and follow the instructions that appear.
Calculating Implicit Queue Limits
This feature imposes, internally, a queue limit on shape queues at ingress and output queues at egress for all queues that are in service. Previously, there was no such queue limit imposed internally, the absence of which creates the possibility of unwanted latencies or buffer starvation during bursts and congestion.
The value of the implicit queue limit is calculated from the following factors, in order of priority:
1. The implicit queue limit is equal to any queue limit already explicitly configured on the queue.
2. If random-detect is configured, the implicit queue limit is equal to double the maximum value of the maximum threshold configured on the queue.
3. The implicit queue limit is equal to 200 milliseconds (ms) of latency at the expected minimum rate of service of the class, assuming an average packet size of 256 bytes.
For example, if the class rate of service is 200 kilobits per second (kbps), the implicit queue limit is as follows:
Implicit_queue_limit_packets = 200,000bits/sec * 200ms * sec/1000ms * packet/256 bytes * bytes/8bits
When computing the implied queue limit, you should be mindful not to oversubscribe the total availability of buffers. Average queue limit is computed for the class, which is the ratio of the rate of this class times the total buffers (256K).
For an Engine 3 Gigabit Ethernet line card, if the rate of the class is 100 Mbps, the average queue limit for this class would be 100M/4G or 2.5 percent of 6553.
In some configurations, as shown in the following policy, it is impossible to determine the service rate of a class.policy fooclass voippriorityclass bbandwidth remaining percent 20class class-default
The rule of thumb in this case is to use a queue limit based on the entire interface bandwidth for the "voip" class, and 50 percent of the interface bandwidth for the "b" and "class-default" classes.
No mechanism exits to disable the Implicit Queue Limit feature other than to override the implicit queue limit with an explict queue limit.
Use the show policy-map interface EXEC command to display queue size statistics.
The following sections provide references related to QoS: Implicit Queue Limit.
Related Topic Document Title
QoS commands: complete command syntax, command modes, command history, defaults, usage guidelines, and examples.
Copyright © 2004 Cisco Systems, Inc. All rights reserved.