Re: [LARTC] HTB and theory

[Stef Coene <stef.coene@docum.org>]

On Monday 09 December 2002 18:14, Catalin Bucur wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hello,
>
> Let's say that my ISP gives me 5000Kbit guaranteed bandwidth. I'm
> starting a HTB traffic shape like this:
>
> tc qdisc add dev eth1 root handle 11: htb  default 99
> tc class add dev eth1 parent 11:0 classid 11:1 htb rate 10000Kbit burst
> ceil 10000Kbit prio 0
> tc class add dev eth1 parent 11:1 classid 11:2 htb rate 5000Kbit ceil
> 5000Kbit prio 5
> [here I have a lot of sub-classes that borrow from parent 11:2]
>
> I'll let HTB to automatically compute the values for 'burst' and
> 'cburst'. The problem is elsewhere. What are the correct values for
> 'rate' and 'ceil' of 11:2 class in this case? In fact, total value of
> 'ceil's from all sub-classes exceeds 5000Kbit, so there are moments when
> the bandwidth that comes from my ISP is bigger than guaranteed bandwidth.
> Is there some kind a theory that says how to establish the values of
> 'rate's and 'ceil's from the parent and its sub-classes?
There are some rules : ceil of child <= ceil of parent, sum (child rates) <= 
rate of parent ....  You don't have to follow this rules, but the final 
shaping result can be strange.
See the faq page on www.docum.org.

Stef

-- 

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

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