I’ve thought of a way I wish the bursting of queues work. That is I would like a queue’s bursting to be controlled no only by what that queue has been doing during the time period of the queue, but I would also like the bursting to be dependent on what the higher priority queues are doing. For example, I would like to be able to set a priority 3 queue to burst if the traffic of the higher priority queues traffic is below a certain level.
Is there any way to accomplish this now as bursting is set up?
why do you need that? if other queue has higher priority, then it will receive its bandwidth at the first place, so your queue can use the rest for bursting - you don’t waste your bandwidth waiting for some ‘levels’
That’s a good point and I was thinking the same thing but there is a difference between the method you’re talking about now and what I’m thinking of and why. The way it’s currently done with the queues and bursting is you specify some level at which traffic must be below before bursting will occur. You could call the waiting that would occur in that situation as well to be wasted.
Perhaps my situation is unique but here’s why my idea would be better for me. Using the currently available method and letting priority levels do the job only works if all the queues in question are subqueues to a parent queue with an overall bandwidth limit. That limiting when effective will end up queuing bytes, apparently even for the highest priority queue. Here I have a crummy satellite internet connection and Skype is our highest priority in the router and also in our thinking. So Skype and DNS are p1 and nothing else is p1. When I use the method you mention, the only one that’s available now, when there is a lot of traffic Skype starts getting a really long delay which is disconcerting. So instead what I’ve done is taken all the other queues which are p2-p8 and put them in a parent queue with the parent having a bandwidth limit to take advantage of the type of limiting you speak of. Skype’s p1 queue is outside of that parent queue which contains all the other queues. So Skype’s queue doesn’t have a bandwidth limit nor does the final parent queue which contains all the queues I’ve mentioned. But because above Skype there are no parent queues with limits I can’t take advantage of the priorities to have Skype take bandwidth from lower queues. So the lower queues are held down to a certain limit leaving some bandwidth for Skype to have. So what I end up with is this. We have a 512k/128k connection. I have the download queues for p2-p8 in a parent queue that’s capped at 400k leaving Skype the rest to have free and clear. Remember I have to because for the final parent queues there’s no limits to invoke the priority level based bandwidth redistribution between queues.
If I could have it the way I want, I could tell the parent queue containing queues p2-p8 to go ahead and burst up to 500k if p1 is below some level.
Does that make sense? If not I’ll try and clarify more.