[altq 691] Re: interesting behavior

[Nobuaki Takahashi <taka@kaynet.ecc.u-tokyo.ac.jp>]

Thanks again, Kenjiro san.


On Thu, 11 Jan 2001 19:53:04 +0900
Kenjiro Cho <kjc@csl.sony.co.jp> wrote:
> You should not make udp_class as default.
> What I meant was something like:
> ===========================================================
> interface xl1 bandwidth 100M cbq
> class cbq xl1 root_class NULL pbandwidth 100
> class cbq xl1 def_class root_class borrow pbandwidth 10 default
> class cbq xl1 udp_class root_class pbandwidth 80
> 	filter xl1 udp_class 192.168.3.2 0 0 0 17
> ===========================================================

I see, I will try this out soon.


> Comparing these outputs, the problem seems to be bursty transmission
> by the sender (coupled with the timer granularity of CBQ).
> The "delays" field indicates that udp_class gets suspended several
> times in the NG case but it is never suspended in the OK case.
> When there's no competing traffic, a burst of packets are transmitted
> and exceeds the assigned rate for a short time period, even though the
> average rate is 30Mbps for a longer time period. 
> When there's competing traffic, a burst becomes smoother since
> competing traffic is interleaved.

Oh thank you very much for analizing the altqstat files.
Now I clearly understand the behavior.
... but I still have one more question about this issue, when 
the data stream suspended in the case like this, do the bursty
packets will be discarded? or kept in the kernel for a while to 
smooth the output data stream?

Regards,
Nobuaki Takahashi

--
  高橋伸明  taka@kaynet.ecc.u-tokyo.ac.jp
  http://www.kaynet.ecc.u-tokyo.ac.jp/~taka/