FortiSwitch QoS with WRR

[lk777]

FortiSwitch 108E

Standalone mode

7.2.3

I have created ip-dscp-map with 4 queues (0-3) which follows Cisco 4-Class QoS Strategy Model.

In qos-policy (egress policy) I selected Weighted Round-Robin Scheduling.

If I understand it correctly, the queues in the qos-policy are those that were assigned in ip-dscp-map.

Are these queues have priority in descending order (queue-7 has the highest priority) only when the Strict Scheduling is used?

When WRR is used, the queue prioritization is determined via weight only. Is this correct?

What do those 0-63 weight values mean? Is this some abstract relative numbers or it has something to do with number of packets, bytes, frames ....?

What weight numbers would fit my 4-Class model:

queue-0 Best Effort (DF- cisco terminology)

queue-1 Mission-Critical Data,Transaction Data, Bulk Data

queue-2 Signalling, Streaming Video

queue-3 Voice, Interactive Video, Routing

Rated By: Percentage (for example):

FS-108E allows only set Max Percent.

Is it percentage of the bandwidth? And it doesn't affect queues prioritization, right?

If this is max bandwidth limiting I think I can ignore it and rely only on the WRR.

Please help me to clarify this subject.

Thanks.

[gfleming]

I think most of your answers are found in the admin guide:

https://docs.fortinet.com/document/fortiswitch/7.2.3/administration-guide/609014/queuing

Please review and let us know if you have further questions. Thanks!

[lk777]

I have posted my question after reading the admin guide. It is not clear to me, that is why I am asking for the forum help.

[gfleming]

---

See references to documentation and explanations for each of your queries:

In qos-policy (egress policy) I selected Weighted Round-Robin Scheduling.

If I understand it correctly, the queues in the qos-policy are those that were assigned in ip-dscp-map.

---

Correct: https://docs.fortinet.com/document/fortiswitch/7.2.3/administration-guide/609011/configuring-fortisw...

Are these queues have priority in descending order (queue-7 has the highest priority) only when the Strict Scheduling is used?

Correct: https://docs.fortinet.com/document/fortiswitch/7.2.3/administration-guide/609014/queuing

When WRR is used, the queue prioritization is determined via weight only. Is this correct?

Correct: https://docs.fortinet.com/document/fortiswitch/7.2.3/administration-guide/609014/queuing

What do those 0-63 weight values mean? Is this some abstract relative numbers or it has something to do with number of packets, bytes, frames ....?

What weight numbers would fit my 4-Class model:

queue-0 Best Effort (DF- cisco terminology)

queue-1 Mission-Critical Data,Transaction Data, Bulk Data

queue-2 Signalling, Streaming Video

queue-3 Voice, Interactive Video, Routing

From the admin guide:

Weighted Round Robin (WRR): Each of the eight egress queues is assigned a weight value ranging from 0 to 63. The purpose of weighted round robin scheduling is to provide prioritized access to the egress port bandwidth, such that queues with higher weight get more of the bandwidth, but lower priority traffic is not starved.

Consider it as a proportional method to service a queue more or less than other given queues. Therefore as an example if you assign a weight of 5, 10, 25, 40, respectively, to your four queues (0 to 3), each queue (assuming they are all full) will be serviced 1/15 (5/75) of the time, 2/15 of the time, 5/15 of the time and 8/15 of the time, respectively.

Rated By: Percentage (for example):

FS-108E allows only set Max Percent.

Is it percentage of the bandwidth? And it doesn't affect queues prioritization, right?

If this is max bandwidth limiting I think I can ignore it and rely only on the WRR.

https://docs.fortinet.com/document/fortiswitch/7.2.3/administration-guide/609011/configuring-fortisw...

From that page, max-rate determines the maximum kbps to assign to a given queue. max-rate-percent assigns a maximum percentage of the ports bandwidth. It does not affect prioritization just maximum bandwidth. It could be useful to set a maximum rate for priority queue so it does not starve other queues.