Policy Manager allows you to configure transmit queues as a component of a class of service (CoS).
There are three transmit queue configuration capabilities:
- Transmit Queue Configuration - Allows you to set the transmit queue associated with the class of service.
- TxQ Shaper - Transmit Queue Rate Shapers let you pace the rate at which traffic is transmitted out of a transmit queue.
- Bandwidth Configuration - Allows you to specify how the traffic in each transmit queue is serviced as it egresses the port.
These three capabilities are configured in the Class of Service Configuration window available from the Policy Manager Edit menu.
For more information, see the section on transmit queues in Getting Started with Class of Service.
Instructions on:
Transmit Queue Configuration
Transmit queues represent the hardware resources for each port that are used in scheduling packets for egressing the device. By default, the static classes of service 0-7 map to transmit queues 0-7. The actual transmit queue number may vary depending on the number of queues supported by the port.
The Queue column in the Class of Service Configuration window displays the actual transmit queues associated with the class of service for each port type. Double-click in the column to see a drop-down menu where you can select a new transmit queue for all port types, or select a different transmit queue for each individual port type.
| TIP: | For more detailed information, refer to the tooltip that appears when you hover the cursor over the Queue column. |
Transmit Queue Bandwidth Configuration
The transmit queue arbiter mode is the method used to determine how traffic in each transmit queue is serviced. It is based on a percentage or weight (called a "slice") given to each queue.
There are three types of arbiter mode: strict priority mode, weighted fair mode, and Enhanced Transmission Selection mode. Each of these modes handles egress traffic differently. In addition, some devices support low latency queues (LLQ) which also impact how egress traffic is handled.
Strict Mode
By default, ports are set to Strict mode, which means that the highest priority queue (the highest numbered queue) is set to 100%. In Strict mode, queues are serviced by numerical priority from the highest numbered queue to the lowest, and all frames in the highest priority queue will be transmitted before the frames in lower priority queues.
Strict priority queuing assures that the highest priority queue with any packets in it will get 100 percent of the bandwidth available. This is particularly useful for one or more priority levels with low bandwidth and low tolerance for delay. The problem with strict priority queuing is that if the higher level queues never fully empty, lower level queues can be starved of bandwidth.
Strict priority queuing is depicted in the following figure. Inbound packets enter on the upper left and proceed to the appropriate queue based upon the TxQ configuration in the CoS. Outbound packets exit the queues on the lower right. In the figure, only queue 3 packets are forwarded, and this will be true until queue 3 is completely empty. Queue 2 packets will then be forwarded. Queue 1 packets will only forward if both queue 2 and queue 3 are empty. Queue 0 packets will only forward if all other queues are empty.
Strict Priority Queuing
Weighted Fair Mode
You can change the arbiter mode to Weighted Fair Queuing, which lets you adjust the slice percentage for each queue and prevent a lower priority queue from being starved. Queues are serviced according to the percentage or weight you assign to each queue. This prevents a lower priority queue from being starved. Percentages must add up to 100%. (Configuring 100% for the highest priority queue sets the port to Strict mode.)
The following figure depicts how weighted fair queuing works. Inbound packets enter on the upper left of the box and proceed to the appropriate priority queue. Outbound packets exit the queues on the lower right. Queue 3 has access to its percentage of time slices so long as there are packets in the queue. Then queue 2 has access to its percentage of time slices, and so on round robin.
Weighted Fair Queuing
Weighted fair queuing assures that each queue will get at least the configured percentage of bandwidth time slices. The value of weighted fair queuing is in its assurance that no queue is starved for bandwidth. The downside of weighted fair queuing is that packets in a high priority queue, with low tolerance for delay, will wait until all other queues have used the time slices available to them before forwarding. So weighted fair queuing would not be appropriate for applications with high sensitivity to delay or jitter, such as VoIP.
Enhanced Transmission Selection Mode
You can change the arbiter mode to Enhanced Transmission Selection (ETS), which allows you to designate 2 or more transmit queues as ETS queues. The ETS queues are then assigned bandwidth allocation with the sum of the ETS queues bandwidth equaling 100%. The scheduler will then service all non-ETS queues first using strict priority. The remaining bandwidth is then distributed based on the allocation that was defined for each of the ETS queues. The priorities within an ETS queue are serviced by strict priority.
The following diagram depicts how ETS works. Priorities 7, 6, and 5 are not part of ETS and will be serviced first. Once these queues have been processed, the remaining bandwidth is then allocated to ETS queues 3 and 0. The ETC queues are then assigned 70 and 30 % bandwidth.
Enhanced Transmission Selection
The Traffic Class (TC) identifiers are first assigned from the lowest to highest based on the priority of the TxQs which the 802.11D maps. This is determined by the Class of Service (CoS) configuration. If the 802.11D priority maps to a customer TxQ (1-8) which is configured for ETS, all priorities which map to the same traffic class groups are assigned the highest priority TC identifier.
| NOTE: | Low Latency Queuing (LLQ) ports may not be assigned to ETS groups. |
|---|
Low Latency Queuing
Extreme Networks K-Series, S-Series, and N-Series devices (with firmware version 7.0.1) support Low Latency Queuing (LLQ) also known as Hybrid queuing. The two highest queues (highest priority) and the lowest queue (lowest priority) will always be designated as LLQ queues. With LLQ, your queuing configuration might look something like:
1) Voice (low-latency, high priority, strict queue)
2) Video (low-latency, high priority, strict queue)
3) Traditional data traffic such as TCP (weighted fair queuing and rate
shaping)
4) Guest traffic or internet traffic (low-latency, low priority, strict queue)
Note that Strict or Weighted Fair Queuing are still used to configure the slice distribution on the non low-latency queues carrying traditional data traffic.
| NOTE: | N-Series Gold devices do not support arbiter mode and slice percentage configuration.
Ports on these devices that are included in the port group will ignore these
settings. |
|---|
Setting the Arbiter Mode
To specify the arbiter mode:
- In the Class of Service Configuration window (available from the Policy Manager Edit menu), double-click in the Transmit Queue Bandwidth column. The Edit Bandwidth Configuration Window window opens.
- Select the desired transmit queue Arbiter Mode: Strict, Weighted Fair Queuing, or Enhanced Transmission Selection. If you want to specify an arbiter mode independently for each port type (for example, 11 Queue Ports or 16 Queue ports), select Use Per-Port Type Arbiter Mode.
- If you have selected Weighted Fair Queuing, click the Edit button to open the Slice Percentages Window. Set the desired slice configuration and click OK.
- Click OK to close the Edit Bandwidth Configuration window.
- Click Save on the toolbar to save the configuration change to the database.
Transmit Queue Rate Shapers
Rate shapers let you pace the rate at which traffic is transmitted out of a transmit queue. Packets received above the configured rate are buffered rather than dropped. Only when the buffer fills are packets dropped.
The following steps describe how to configure rate shapers in Policy Manager:
- In the Class of Service Configuration window (available from the Policy Manager Edit menu), select the class of service where you want to configure the transmit queue.
- Double-click on the selected line below the TxQ Shaper column and select the desired rate shaper from the drop-down menu. For information on how to create a new rate, see the Create Rate Limit/Shaper Window Help topic.
- Click Save on the toolbar to save the configuration change to the database.
For more information, see the section on transmit queues in Getting Started with Class of Service.
| NOTE: | A rate shaper is associated to a specific transmit queue, not a CoS. This means that the 1) you should select the queue you want to use for a CoS first, then set the shaper and 2) all CoS using that queue will use the same rate shaper. |
For information on related concepts:
For information on related tasks:
