[c-nsp] Output Drops Due to QoS and threshold size

[Victor Sudakov]

Saku Ytti wrote:
> 
> > My question is, what's the tradeoff? Why is the default threshold size
> > the mere 100 and not 3100? What will happen if I make it 3100 on all
> > queues in all the switches and forget about drops forever? Is it
> > dangerous?
> 
> I'd recommend reading this through:
> https://supportforums.cisco.com/document/31581/egress-qos

Thank you for the link. It has taken me several days to digest the
somewhat lengthy document (little by little), and it contains some
really useful and helpful information.

However, all the solutions offered there (the "Quick and Easy Egress
Buffer Optimizing Solution" and the more elaborate solutions with
different classes of traffic being mapped to different queues and
thresholds) are ALL based on the same principle: increasing the
relevant thresholds to 1000% and more.

My question still holds: why are the default thresholds so low? This
must mean something? What will happen if I make them 3000 or 2000 on
all queues in all the interfaces (or the uplink interfaces) and forget
about drops forever?  What do I lose or risk? 

Let me quote "3.1 Quick and Easy Egress Buffer Optimizing Solution"

	This is for those seeking a minor tweak to fix a problem with the
	occasional egress packet drop. The quick answer is to increase the
	drop thresholds used by WTD.

	Switch(config)#mls qos queue-set output 1 threshold 1 1000 1000 50 1000
	Switch(config)#mls qos queue-set output 1 threshold 2 1000 1000 50 1000
	Switch(config)#mls qos queue-set output 1 threshold 3 1000 1000 50 1000
	Switch(config)#mls qos queue-set output 1 threshold 4 1000 1000 50 1000

Why is it *not* the default configuration?

-- 
Victor Sudakov,  VAS4-RIPE, VAS47-RIPN
sip:sudakov at sibptus.tomsk.ru