Re: ingress policing based on source address?

Linux Advanced Routing and Traffic Control

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Simon Oosthoek wrote:
Hi all

I'm new to this list, but not exactly to iproute stuff.

I'd like to solve a specific problem with bandwidth coming from
different external sources towards the internal network (also the other
way around, but I figure that's not so much a problem, since that is
egress traffic shaping).

The network looks like this:

internet ------ ISP-------[shaping/router]
 |                |                +-   net1 -------- host1
mirrors          host2              +-   net2

in text: we connect to the internet via an ISP, where we also have an
externally accessible host (host2). Internally we use NAT and several
subnets.

We have a 100Mbit/s connection to the ISP, but we only pay for 1Mbit/s.
So in order to keep our traffic within the agreed parameters, we need to
police our incoming and outgoing traffic. However the traffic from and
to the ISP and host2 doesn't have to be policed.

For our external traffic there's not much problem to shape the traffic
in the egress queues (using HTB and TBF/SFQ stuff). This is well
described in the LARTC howto documentation.

My problem is with the incoming traffic. The examples in the howto don't
go very much into this and from what I understand the ingress queue is
much less advanced than the egress queue. I read something about an
intermediate queue, but I don't understand how that works (yet).

My question is whether this is something I can do using the ingress
queue, somehow defining filter rules with different queue associated
with that, or whether someone has experience with a similar configuration.

If this is not possible, I might be able to solve it differently by
adding another host with just 2 interfaces and using only egress queues,
but I'd prefer limiting my solution to a single host.

You could use IMQ www.linuximq.net to save you using another box, but the way you get your bandwidth (ie. no real bottleneck) may make it suitable for just the "dumb" ingress policer. Maybe you have good reasons to want to classify/shape ingress in a complicated way.


Andy.

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

[Index of Archives]     [LARTC Home Page]     [Netfilter]     [Netfilter Development]     [Network Development]     [Bugtraq]     [GCC Help]     [Yosemite News]     [Linux Kernel]     [Fedora Users]
  Powered by Linux