Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 27 Apr 2012 15:02:00 +0200
From:      Luigi Rizzo <rizzo@iet.unipi.it>
To:        Javier - <javier_9185@hotmail.com>
Cc:        freebsd-ipfw@freebsd.org
Subject:   Re: Dummynet and bursting!
Message-ID:  <20120427130200.GA17737@onelab2.iet.unipi.it>
In-Reply-To: <SNT136-W154D1068D52BC93BA4553D96250@phx.gbl>
References:  <20120427065523.GA90180@onelab2.iet.unipi.it> <SNT136-W154D1068D52BC93BA4553D96250@phx.gbl>

next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Apr 27, 2012 at 12:40:05PM +0000, Javier - wrote:
> 
> OK, but with increased burst to 5mbytes i have same results.
> 

the issue is the bandwidth, not the burst.
it is possible that the system has a bottleneck
similar to the 125k you are configuring.
Besides, the tcp window or socket buffer might be
small (say 64Kbytes) so you won't be able to
use the burst anyways.

luigi

> will try fresh install from freebsd (no pfsense), and the results will be posted later.
> 
>  
> 
> Thanks!
> 
>  
> 
> ---------------------------------------------------------------
> 
> Transfer estimates are completely unreliable in general,
> you need time for tcp to open the window, plus the RTT
> of the connection influences the throughput, etc.
>  
> You can perhaps make some reasonable test by setting a low
> bandwidth. Burst seems to work fine here (on stable/9), see below
>  
> --- no pipes, no nothing ---
>  > ping localhost
>  PING localhost (127.0.0.1): 56 data bytes
>  64 bytes from 127.0.0.1: icmp_seq=0 ttl=64 time=0.079 ms
>  64 bytes from 127.0.0.1: icmp_seq=1 ttl=64 time=0.064 ms
>  ^C
>  
> --- one 10Kbit/s pipe, no burst ---
>  > sudo ipfw add 100 pipe 1 proto icmp out
>  00100 pipe 1 ip from any to any proto icmp out
>  > sudo ipfw pipe 1 config bw 10Kbit/s burst 0
>  > sudo ipfw pipe show
>  00001: 10.000 Kbit/s 0 ms burst 0 
> q131073 50 sl. 0 flows (1 buckets) sched 65537 weight 0 lmax 0 pri 0 droptail
>  sched 65537 type FIFO flags 0x0 0 buckets 0 active
>  > ping localhost
>  PING localhost (127.0.0.1): 56 data bytes
>  64 bytes from 127.0.0.1: icmp_seq=0 ttl=64 time=134.602 ms
>  64 bytes from 127.0.0.1: icmp_seq=1 ttl=64 time=135.011 ms
>  64 bytes from 127.0.0.1: icmp_seq=2 ttl=64 time=135.019 ms
>  ^C
>  
> --- now increase the burst to 100Kbytes ---
>  > sudo ipfw pipe 1 config bw 10Kbit/s burst 100K
>  > sudo ipfw pipe show
>  00001: 10.000 Kbit/s 0 ms burst 102400 
> q131073 50 sl. 0 flows (1 buckets) sched 65537 weight 0 lmax 0 pri 0 droptail
>  sched 65537 type FIFO flags 0x0 0 buckets 0 active
>  > ping localhost
>  PING localhost (127.0.0.1): 56 data bytes
>  64 bytes from 127.0.0.1: icmp_seq=0 ttl=64 time=0.070 ms
>  64 bytes from 127.0.0.1: icmp_seq=1 ttl=64 time=0.083 ms
>  64 bytes from 127.0.0.1: icmp_seq=2 ttl=64 time=0.066 ms
>  ^C
>  
> cheers
> luigi
> ----------------------------------------
> > Date: Fri, 27 Apr 2012 13:49:41 +0200
> > From: rizzo@iet.unipi.it
> > To: javier_9185@hotmail.com
> > CC: freebsd-ipfw@freebsd.org
> > Subject: Re: Dummynet and bursting!
> >
> > On Fri, Apr 27, 2012 at 10:50:17AM +0000, Javier - wrote:
> > >
> > > I want to leave at cable speed n bytes, after n bytes apply the queue bw limit...
> >
> > and what are you seeing instead ? Do you have a trace or
> > something that shows that it does not work like this ?
> >
> > cheers
> > luigi
> >
> > > In Linux with htb this is done with cburst parameter...
> > >
> > >
> > >
> > > Regards!
> > >
> > >
> > > > Date: Fri, 27 Apr 2012 08:55:23 +0200
> > > > From: rizzo@iet.unipi.it
> > > > To: javier_9185@hotmail.com
> > > > CC: freebsd-ipfw@freebsd.org
> > > > Subject: Re: Dummynet and bursting!
> > > >
> > > > On Fri, Apr 27, 2012 at 04:57:20AM +0000, Javier - wrote:
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > Hello all!, i?m researching about bursting with dummynet?s pipes. i can?t find the way in order to make the burst paramter work under pipe command;pipe config ........ burst 500000 (500Kbytes)->not luck pipe config ........ burst 500000 queue 500Kbytes ->not luck These parameters seems to be ok under pipe show, but no effect on real escenario... traffic is anyway at same speed defiined by "bw". Any help/comment wll be appreciated...specially from Luigi! Is he alive on the list? Regards! _______________________________________________
> > > >
> > > > can you clarify what you are expecting and what you are seeing ?
> > > >
> > > > cheers
> > > > luigi 		 	   		  



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20120427130200.GA17737>