* [LARTC] Guaranteed bandwidth per connection
@ 2004-04-21 9:25 Jeroen Vriesman
2004-04-21 13:54 ` Radoslav Kolev
2004-04-21 14:42 ` Jeroen Vriesman
0 siblings, 2 replies; 3+ messages in thread
From: Jeroen Vriesman @ 2004-04-21 9:25 UTC (permalink / raw)
To: lartc
Dear all,
I've got a working HTB configuration with iptables, fwmark, SFQ etc.
At the moment, I can mark traffic and give it a maximum bandwidth and a minimum guaranteed bandwidth, so far so good.
What I would like to do is the following:
In stead of defining a min/max for a certain type of traffic (e.g. http, ftp whatever), I would like to define a "minimum guaranteed bandwidth per connection".
e.g. An application connecting to port X would get 10kbit/s guaranteed, the next connection to port X would also get 10kbit/s etc.
Would be something like having N (the maximum number of connections) HTB classes, and put every new connection in another class.
Does anyone know how to do that?
Kind regards,
Jeroen Vriesman.
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [LARTC] Guaranteed bandwidth per connection
2004-04-21 9:25 [LARTC] Guaranteed bandwidth per connection Jeroen Vriesman
@ 2004-04-21 13:54 ` Radoslav Kolev
2004-04-21 14:42 ` Jeroen Vriesman
1 sibling, 0 replies; 3+ messages in thread
From: Radoslav Kolev @ 2004-04-21 13:54 UTC (permalink / raw)
To: lartc
Hi!Jeroen Vriesman wrote:
>In stead of defining a min/max for a certain type of traffic (e.g. http, ftp whatever), I would like to define a "minimum guaranteed bandwidth per connection".
>e.g. An application connecting to port X would get 10kbit/s guaranteed, the next connection to port X would also get 10kbit/s etc.
>
>Would be something like having N (the maximum number of connections) HTB classes, and put every new connection in another class.
>
>
I don't know how you can put every new connection in a new class, but
why don't you try this:
Decide what will be the maximum number of connections you want to
provide the guaranteed minimum and how much will it be (kbit/s).
If for example we take connections to port 80, create a HTB class with
rate=max_conn*guaranteed_rate and classify all traffic to port 80 to it.
Then add an SFQ behind. This way if there is a fewer number of
connections than you planned for each will get
(class_rate/connections)kbit, which
will more than the guaranteed minimun. If the number is equal to the
number of connections, the SFQ will make sure that no particular connection
take over the bandwitdh, and evey connection will get it's guaranteed
minimum. If there are more connections than you expected, each will get
an equal portion of the available bandwidth. Actually that's exactly
what SFQ is designed for, to distribute the available bandwidth equally
to every connection.
Greets,
Rado
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [LARTC] Guaranteed bandwidth per connection
2004-04-21 9:25 [LARTC] Guaranteed bandwidth per connection Jeroen Vriesman
2004-04-21 13:54 ` Radoslav Kolev
@ 2004-04-21 14:42 ` Jeroen Vriesman
1 sibling, 0 replies; 3+ messages in thread
From: Jeroen Vriesman @ 2004-04-21 14:42 UTC (permalink / raw)
To: lartc
Hi,
well, that's exactly what I'm doing now.
The only thing is that it only makes sense if there are a lot of connections, when there are only a few, they will get a lot of bandwidth, which isn't what I want.
Thanks for the response.
Jeroen.
On Wed, 21 Apr 2004 16:54:11 +0300
Radoslav Kolev <radolin@del.bg> wrote:
> Hi!Jeroen Vriesman wrote:
>
> >In stead of defining a min/max for a certain type of traffic (e.g. http, ftp whatever), I would like to define a "minimum guaranteed bandwidth per connection".
> >e.g. An application connecting to port X would get 10kbit/s guaranteed, the next connection to port X would also get 10kbit/s etc.
> >
> >Would be something like having N (the maximum number of connections) HTB classes, and put every new connection in another class.
> >
> >
> I don't know how you can put every new connection in a new class, but
> why don't you try this:
> Decide what will be the maximum number of connections you want to
> provide the guaranteed minimum and how much will it be (kbit/s).
> If for example we take connections to port 80, create a HTB class with
> rate=max_conn*guaranteed_rate and classify all traffic to port 80 to it.
> Then add an SFQ behind. This way if there is a fewer number of
> connections than you planned for each will get
> (class_rate/connections)kbit, which
> will more than the guaranteed minimun. If the number is equal to the
> number of connections, the SFQ will make sure that no particular connection
> take over the bandwitdh, and evey connection will get it's guaranteed
> minimum. If there are more connections than you expected, each will get
> an equal portion of the available bandwidth. Actually that's exactly
> what SFQ is designed for, to distribute the available bandwidth equally
> to every connection.
>
> Greets,
> Rado
>
> _______________________________________________
> LARTC mailing list / LARTC@mailman.ds9a.nl
> http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
>
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2004-04-21 14:42 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-04-21 9:25 [LARTC] Guaranteed bandwidth per connection Jeroen Vriesman
2004-04-21 13:54 ` Radoslav Kolev
2004-04-21 14:42 ` Jeroen Vriesman
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.