All of lore.kernel.org
 help / color / mirror / Atom feed
* [LARTC] Excess BW again
@ 2003-10-15 10:42 Tom Olexa
  2003-10-16 17:29 ` Stef Coene
  0 siblings, 1 reply; 2+ messages in thread
From: Tom Olexa @ 2003-10-15 10:42 UTC (permalink / raw)
  To: lartc

Hello there,

I've already posted st similar, but anyway...
can anyone tell me, why the following setup

tc qdisc add dev eth0 root handle 1: htb r2q 5 

tc class add dev eth0 parent 1: classid 1:1 htb rate 512kbit ceil
512kbit tc class add dev eth0 parent 1:1 classid 1:10 htb rate
64kbit ceil 512kbit tc class add dev eth0 parent 1:1 classid 1:11
htb rate 256kbit ceil 512kbit

tc filter add dev eth0 parent 1: protocol ip prio 1 u32 \
        match ip dst 195.28.103.7 flowid 1:10
tc filter add dev eth0 parent 1: protocol ip prio 1 u32 \
        match ip dst 195.28.103.5 flowid 1:11

doesn't distribute the excess bw in proportion of allocated bw?

Ratio I expect is 64/256, but I get some 1/2.1 or similar. This
varies according to r2q, but anyway it never gets close to 1/4.

kernel 2.4.21, htb 3.12, but also tested on 2.4.18, htb 3.6 with
similar results...

Can anyone try this setup?

Thanks, Tom

-- 
Potrebujete vice prostoru pro vase stranky?
Ptejte se na http://sluzby.volny.cz/cs/product/ftp_paid


_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [LARTC] Excess BW again
  2003-10-15 10:42 [LARTC] Excess BW again Tom Olexa
@ 2003-10-16 17:29 ` Stef Coene
  0 siblings, 0 replies; 2+ messages in thread
From: Stef Coene @ 2003-10-16 17:29 UTC (permalink / raw)
  To: lartc

On Wednesday 15 October 2003 12:42, Tom Olexa wrote:
> Hello there,
>
> I've already posted st similar, but anyway...
> can anyone tell me, why the following setup
>
> tc qdisc add dev eth0 root handle 1: htb r2q 5
>
> tc class add dev eth0 parent 1: classid 1:1 htb rate 512kbit ceil
> 512kbit tc class add dev eth0 parent 1:1 classid 1:10 htb rate
> 64kbit ceil 512kbit tc class add dev eth0 parent 1:1 classid 1:11
> htb rate 256kbit ceil 512kbit
>
> tc filter add dev eth0 parent 1: protocol ip prio 1 u32 \
>         match ip dst 195.28.103.7 flowid 1:10
> tc filter add dev eth0 parent 1: protocol ip prio 1 u32 \
>         match ip dst 195.28.103.5 flowid 1:11
>
> doesn't distribute the excess bw in proportion of allocated bw?
>
> Ratio I expect is 64/256, but I get some 1/2.1 or similar. This
> varies according to r2q, but anyway it never gets close to 1/4.
>
> kernel 2.4.21, htb 3.12, but also tested on 2.4.18, htb 3.6 with
> similar results...
>
> Can anyone try this setup?
I tried it myself and this is the result (in bytes and the output is not like 
it should be):

2.884 KB/s 4.258 KB/s  T 7.143 KB/s  G 75.41 KB/s 40.38%  40.38%
18.66 KB/s 41.64 KB/s  T 60.31 KB/s  G 67.82 KB/s 30.95%  36.17%
21.54 KB/s 43.08 KB/s  T 64.62 KB/s  G 66.75 KB/s 33.33%  35.25%
22.97 KB/s 45.95 KB/s  T 68.92 KB/s  G 67.30 KB/s 33.33%  34.76%
21.54 KB/s 44.52 KB/s  T 66.06 KB/s  G 67.05 KB/s 32.61%  34.33%
21.54 KB/s 43.08 KB/s  T 64.62 KB/s  G 66.64 KB/s 33.33%  34.17%
21.54 KB/s 44.51 KB/s  T 66.06 KB/s  G 66.56 KB/s 32.61%  33.95%
21.53 KB/s 43.07 KB/s  T 64.61 KB/s  G 66.32 KB/s 33.33%  33.88%
24.41 KB/s 43.07 KB/s  T 67.48 KB/s  G 66.45 KB/s 36.17%  34.13%
17.23 KB/s 44.53 KB/s  T 61.76 KB/s  G 65.98 KB/s 27.91%  33.55%
22.97 KB/s 41.64 KB/s  T 64.62 KB/s  G 65.85 KB/s 35.56%  33.73%
21.54 KB/s 44.51 KB/s  T 66.06 KB/s  G 65.87 KB/s 32.61%  33.64%
24.41 KB/s 44.51 KB/s  T 68.93 KB/s  G 66.11 KB/s 35.42%  33.78%
17.23 KB/s 41.64 KB/s  T 58.87 KB/s  G 65.59 KB/s 29.27%  33.49%
22.97 KB/s 44.52 KB/s  T 67.50 KB/s  G 65.72 KB/s 34.04%  33.53%
20.09 KB/s 43.06 KB/s  T 63.16 KB/s  G 65.56 KB/s 31.82%  33.42%
24.42 KB/s 44.54 KB/s  T 68.96 KB/s  G 65.76 KB/s 35.42%  33.55%
18.66 KB/s 43.08 KB/s  T 61.74 KB/s  G 65.53 KB/s 30.23%  33.37%
21.50 KB/s 41.58 KB/s  T 63.09 KB/s  G 65.40 KB/s 34.09%  33.41%
21.57 KB/s 44.58 KB/s  T 66.16 KB/s  G 65.44 KB/s 32.61%  33.37%
12.92 KB/s 43.08 KB/s  T 56.00 KB/s  G 64.99 KB/s 23.08%  32.95%
15.79 KB/s 43.07 KB/s  T 58.87 KB/s  G 64.71 KB/s 26.83%  32.69%
18.67 KB/s 45.95 KB/s  T 64.62 KB/s  G 64.71 KB/s 28.89%  32.53%
(10 rate) (11 rate) (10 + 11) (average 10+11) (10/11) (average 10/11)

So the ratio is 32.53%.
Ideally, this should be 64/256 = 25%
Not that bad.

Stef

-- 
stef.coene@docum.org
 "Using Linux as bandwidth manager"
     http://www.docum.org/
     #lartc @ irc.openprojects.net

_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2003-10-16 17:29 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-10-15 10:42 [LARTC] Excess BW again Tom Olexa
2003-10-16 17:29 ` Stef Coene

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.