Trygve Selmer wrote:
> Address BOOK wrote:
>> Hi all,
>>
>> Due to my unsuccessful attempts,
>>
>> could someone brief me on the Linux Traffic Shaper,
>> Either CBQ- or HTB-based. (Not the 'shaper' or 'MID-something' device.)
>>
>> And help me solve this problem :
>>
>> On my LAN ( 192.168.2.0/24 ), I'd like configure my Linux Box in order
>> to throttle the traffic that goes from the Internet down do my Linux
>> box; that is, through my gateway ( 192.168.2.1 ). Your help will be very
>> appreciated.
>>
>> Only volatile test configuration matters; no init script.
>
> AFAIK, you can only shape outgoing traffic. This means you have to do
> the shaping on your gateway. If eth0 is your Internet and eth1 your lan
> connections, you shape traffic from your workstation to the Internet on
> eth0 and traffic from Internet to your workstation on eth1.
>
Thanks for your answer, but I had good reasons to do so.
What I want is :
1) To have my work station not to consume to much from the Internet
bandwidth so that other stations on my LAN could share together a
guaranteed bandwidth of say 50kbit/s.
2) The way I would like to implement it, since my gateway is quite
stubborn, is by making my workstation to drop incoming TCP packets above
some rate, let's say 150kbit/s, regardless of the connexion they belong
to. (An overall inbound TCP traffic rate limit of 150kbit/s.)
3) This way, my TCP download streams (obviously not UDP things) would
auto-regulate themselves in order not to exceed this barrier, thus
controlling the overall (TCP) consumption of my workstation; which
should represent most of my Internet use, though.
4) I would agree if, for some reason, I had to set up a buffer (queue)
for this incoming traffic.
5) As a reminder, I only want to regulate the inbound TCP traffic
between the Internet and my workstation, not with other stations on my LAN.
6) I must concede that I would have better not to filter on IP traffic
in general, but on TCP traffic only.
In hope I was clear enough,
I thank you for more suggestions.
Bye,
A.B.
|