Getting in touch with Burst Mode - basic understanding

Hi,

I was hoping you could help me on understanding the burst mode in the queue tree.

actual-rate: current bandwidth speed
burst-limit: max speed
burst-threshold: on/off mark

Whenever having a short stream i.e. opening a website, the actual-rate fires up to the burst-limit for a few seconds. Then the burst will not be allowed as long as the average-rate is above the burst-threshold.

So when you have a bigger stream - let’s say you are downloading a software package - and the download rate is reaching its max-limit the whole time after the short burst, you won’t be able to achieve another burst until the stream ends and the actual-rate is getting lower, so the average-rate is getting below the burst-threshold.

If you are setting the burst-threshold above the max-limit the actual-rate will look like a sine wave and the burst will be allowed every few seconds depending on your average-rate.

Is this correct? Did I miss something?

Best wishes,
McShadow

interesting topic :+1:t2:

but unfortunately this qos/sla kind of questions can’t take a short answer.

key points :

  • bandwidth available on interface (or on the network)
  • multipliers (idle/peak time, numbers of concurrent access etc)

which then creates baseline average, which sometimes could be below nor above it.

  • the lower delivery parts, as results of concurrent access, will be accumulated to be delivered when there are spare available bandwidth on the interface/network.

  • the upper delivery parts, as results of the accumulated minus parts of the average bandwidth stated. when those accumulated value filled (as burst amount) - the delivery must be returned back to average stated.

so, those are the short - simplified answers.

hope this helps.