View Full Version : Traffic Shaping
jkempf74
11-08-2009, 11:54 PM
Real quick quuestion for traffic shapping the incoming traffic burst when that is set how long is that good for in the software. If I set it to a 1.5 meg burst does that gice the user a 1.5 meg increase all the time or just in certain circumstances and such. Just looking for some guidance on those options.
UBNT-Mike.Ford
11-09-2009, 12:50 PM
Real quick quuestion for traffic shapping the incoming traffic burst when that is set how long is that good for in the software. If I set it to a 1.5 meg burst does that gice the user a 1.5 meg increase all the time or just in certain circumstances and such. Just looking for some guidance on those options.
Hello,
After every idle period it will allow bursting to 1.5mbps, then back down to the set throughput. If the throughput is not sustained (like web traffic) it will allow for bursting periodically. For file downloads it will still cap at the preset static throughput rate.
Thanks,
jkempf74
11-09-2009, 02:55 PM
How long is the idle period ty
konrad
11-09-2009, 03:52 PM
This might help you: http://en.wikipedia.org/wiki/Token_bucket
jkempf74
11-09-2009, 04:14 PM
That was very helpful ty for the link. What benfits will be had if we give a customer lets say 2.5 meg burst on a 1.5 meg account.
konrad
11-10-2009, 05:07 AM
That really depends and its not easy to answer.
Basically, if the clients traffic will be bursty in nature he will see it as an improvement, if thats 2.5 MB were talking about in this case, if his major traffic is http he will be seeing almost wire speeds (or air speeds in this case).
Unless there are other bottlenecks, be it natural or created by you (like shaping on your isp gateway).
In my opinion a large burst is not exactly a good thing, if thats all shaping you would be doing, it could be (but not that large) if thats just a precaution and you're shaping him elsewhere. Simple TBF discussed here has drawbacks, as it doesn't classify the traffic, so, if the burst has a special purpose (like http acceleration - thats my case), the results may be not what you expect ;)