While it might seem weird, it is a valid scenario. There might be reasons
to let reusing IP addresses.
To answer your question about deployment of NAT devices, it shouldn't be
just a NAT device. Each LAN segment can have its own NAT gateway.
On Wed, Feb 22, 2017 at 4:07 PM, /dev/rob0 <r...@gmx.co.uk> wrote:
> On Wed, Feb 22, 2017 at 02:55:30PM +0100, S. M. Hossein Hamidi wrote:
> > I am new to dnsmasq and my mind is busy with a DHCP scenario that I
> > would like to consult with you before getting my hands dirty.
> > Suppose that we have several subnets which are behind a NAT device
> > which additionally acts as a DHCP relay agent. I know that using
> > DHCP relay agent information option, I can distinguish between
> > different subnets. However, the additional constraint is to use
> > duplicate or overlapping subnets instead of distinct subnets.
> That is a very strange constraint. RFC 1918 is quite large enough;
> it's not necessary to share your netblocks. And how will your NAT
> device distinguish one segment with a shared netblock from another
> segment using the same netblock?
> Sounds like broken IP networking to me. Compliant IP stacks do
> routing based on IP address blocks.
> > Since, the traffic comes from each segment goes through NAT, it
> > wouldn't be any problem to use duplicate IP addresses, knowing
> > that each segment has its own autonomy.
> But again, how does the NAT device do this?
> > So, may I know if can I exploit tagging feature of dnsmasq so
> > that it can handle separate potentially overlapping leases for
> > each subnet?
> I'm pretty sure this would not be possible in dnsmasq nor in ISC
> dhcpd. Either one could do a single netblock shared among many
> distinct segments, but each would only be able to give out any one
> address to only one client.
> But wait and see what Simon says. :)
> Offlist GMX mail is seen only if "/dev/rob0" is in the Subject:
> Dnsmasq-discuss mailing list
Dnsmasq-discuss mailing list