[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ipcop-user
Subject:    Re: [IPCop-user] even more QOS fun
From:       johnny <sk8rdie42 () gmail ! com>
Date:       2009-12-26 23:44:52
Message-ID: 4B369FF4.2000702 () gmail ! com
[Download RAW message or body]


>
> QOS for my little cable modem and reasonably fast Pentium III system, I didn't 
> notice any problems. I tried the same thing at a client site with a high-speed 
> cable connection and, with QOS, the download speed was somewhere between 
> 4000-7000 kb per second. Without QOS, it was 31,000 kb per second. The QOS limit 
> was set at 32,000 kb per second.
>
> Whoops.
>   
FWIW I have never had luck using IPCop's built in traffic shaping for 
cable connections over 10MBit. Unfortunately I discovered Markus 
Hoffman's layer7 filtering and QoS-ng packages for IPCop thus preventing 
me the necessity to dig in to IPcop's code more closely. The MH-addon 
packages previously included the imq kernel module for IPCop. I used 
that for older IPCop versions, and it worked swimmingly. Eventually I 
felt I had to upgrade IPCop, but by then I had learned enough to follow 
simple instructions and patch and install the kernel module for later 
kernel version IPcop versions (not a typo.... I don't believe that is a 
redundant statement).

Anyway, I'm not currently inclined to start digging through IPcop 1.4 
traffic shaping code (maybe after I send this), but I'm sure that the 
LARTC and tc documentation provide plenty of background to be able to 
dig through the IPcop code and answer the simple 'why' of what my 
experience has been using 10+ Mbit cable connections. I have a pretty 
good feeling this may have been discussed in detail in other threads 
anyway (too lazy to search right now).

The only reason I am even mentioning this is that I have seen identical 
behavior from a d-link device that was purchased in 2003 and originally 
used (without issue) with a 6Mbit cable connection. When used with a 10+ 
Mbit connection the speeds never topped 8Mbit. Switching to a newer 
Linksys device yielded 20+ Mbit speeds. This experience between the 
older d-link and newer Linksys reflected my experience using IPcop with 
integrated traffic shaping enabled versus disabled. I have no idea if 
the d-link device was attempting any traffic shaping.

Take it or leave it, my experience has been that the intermediate 
queuing device (imq) is the only reliable way to shape traffic in both 
directions, and that IPcop's traffic shaping works best for getting low 
latency priority traffic from highly utilized DSL speed types of 
connections. For highly asynchronous connections like cable at 10_ Mbit 
down, imq is the better way to go.

Johnny

------------------------------------------------------------------------------
This SF.Net email is sponsored by the Verizon Developer Community
Take advantage of Verizon's best-in-class app development support
A streamlined, 14 day to market process makes app distribution fast and easy
Join now and get one step closer to millions of Verizon customers
http://p.sf.net/sfu/verizon-dev2dev 
_______________________________________________
IPCop-user mailing list
IPCop-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ipcop-user
[prev in list] [next in list] [prev in thread] [next in thread] 

Configure | About | News | Add a list | Sponsored by KoreLogic