|
•
|
The Shaper defines ten traffic classes, four of which are prescriptively named --- real-time, interactive, default, and best effort.
|
|
•
|
The system applies these QoS settings globally after compressing (deduplicating) all the outbound tunnelized and pass-through-shaped traffic --- shaping it as it exits to the WAN.
|
The Configuration - Shaper page looks like this:
 |
What happens if you change the Minimum Bandwidth values?
|
•
|
If all minimums are equal to 0%, then Excess Weighting alone determines bandwidth allocation and no traffic class has priority. (Referred to as pure weights.)
|
Tip 
When you set a traffic class Minimum Bandwidth to zero, you are explicitly not guaranteeing any bandwidth for that class.
When you set a traffic class Minimum Bandwidth to zero, you are explicitly not guaranteeing any bandwidth for that class.|
•
|
If the sum of the percentages for the queues in use exceeds 100%, then low-priority traffic classes might not receive their guaranteed bandwidth (starvation).
|
|
•
|
If all minimums are equal to 100%, then Priorities alone determine bandwidth allocation. For example, Priority 2 only gets bandwidth if Priority 1 is completely satisfied. (Referred to as pure priorities.)
|
How is Excess Weighting calculated and applied?
|
•
|
Excess Weighting is a ratio of the weight of one traffic class divided by the sum of the weights of the active traffic classes. So, if all three traffic classes were active in the example above, Traffic Class 2 would get 1000/(100 + 1000 + 1000) = 1000/2100 = 48% of the excess bandwidth.
|
|
•
|
When you configure the Max WAN Bandwidth, you need to consider two things:
 |
If you set the Max WAN Bandwidth too low, you may underutilize your links. If you set it too high (oversubscribe), you may overrun the appliance WAN link, or cause congestion and drops on the router.
You can enter this value on either the Configuration - Deployment page or the Configuration - Shaper page.
|
•
|
|
•
|
|
•
|
|
•
|
By default, the values are the same for Max WAN Bandwidth (for tunnelized traffic) and the Max Bandwidth for pass-through shaped traffic.
However, you can cap the maximum amount of bandwidth allocated to pass-through shaped traffic by configuring the upper limit at the bottom of the Configuration - Shaper page.
 |
It’s important to note that this is not the same as configuring a percentage of Max WAN BW. This calculation is done after exiting the Shaper, so until that point, all shaped packets have queued through the traffic classes as they arrived. As a result, pass-through packets in a higher priority traffic class have a better chance of getting through in the event that the max is exceeded, or if congestion occurs.
Even though it appears on the Configuration - Tunnels page, as opposed to the Configuration - Shaper page, the Shaper uses the Tunnel Max BW value as it services queues.
 |
After the Max Bandwidth has been met for a given tunnel, the Shaper won’t schedule any more packets for transmission in that tunnel until more bandwidth is available. Since the clock is still ticking for any packets still in a queue for that tunnel, the traffic class Max Wait Time could be exceeded for those packets before bandwidth is available.
Each model of appliance has a specific maximum system bandwidth. That is, the amount of bandwidth it can support for optimized traffic at the WAN interface.
By default, all tunnels are set to automatically negotiate tunnel bandwidth to the lowest common value. The following illustrations show this negotiation from the perspective of an NX-8500 with multiple tunnels. The maximum values assume that all options are enabled.
 |
After negotiating bandwidth for all four tunnels, 119 Mbps (1000 minus 881) are left over for shaped pass-through traffic.
Auto BW can only negotiate the link between two appliances — A and Hub, and B and Hub. So, here it can negotiate the link down to 100 Mbps. However, if A and B both transmit at 100 Mbps, the hub will be overrun.
Enabling Dynamic Rate Control on the Hub allows it to control the tunnel traffic by lowering each remote appliance’s Tunnel Max Bandwidth. The smallest possible value for A or B is that appliance’s Tunnel Min Bandwidth.
 |
