Matt’s excellent Bristol Wireless network diagram
Unfortunately, Matt’s excellent network diagram has disappeared into the ether since this article was first published in October 2003.
Our sincere apologies. 🙂
Unfortunately, Matt’s excellent network diagram has disappeared into the ether since this article was first published in October 2003.
Our sincere apologies. 🙂
The link to …network.gif above appears to be broken. The image and text here is extracted from …network.zip in the same location (and the image has been converted from .bmp to .png). I have no idea how current this is. Caveat lector!
each wireless node has 2 interfaces, eth0 and wlan0.
eth0 is set to 10.21.x.1, x being the node number.
wlan0 is set to 10.20.x.1, x being the node number.
bw-ecc is the first node on the network so x=0, bw-example (see below) would be x=1, and the next node x=2 and so on.
the wlan0 network is split into 2 subnets, the first subnet being used for static address space and client dhcp address space the second subnet being used for wds addressing which are assigned by a dhcp like dynamic allocation program called aladin (written by mike saywell from sown for this very purpose).
bw-ecc:
wlan0: 10.20.0.1
eth0: 10.21.0.1
bw-ecc’s dhcp range for client use on wifi side is 10.20.0.20 to 10.20.0.126
bw-example:
wlan0: 10.20.1.1
eth0: 10.21.1.1
bw-example’s dhcp range for client use on wifi side is 10.20.1.20 to 10.20.1.126
The nodes share kernel routing tables using a protocol called ospf (we use the zebra ospf daemon) which broadcast their routing tables multicast and every node picks them up and corrects its kernel routing table so it knows where to go for each network.
Also worth bearing in mind that Andy Laurence is the custodian of the Consume.net IP allocation for Bristol: 10.12.0.0/16
Should we be using addresses in this range?
I think the original idea was to use the Consume IP ranges, and I did sort out a schema for allocating the addresses across the city. I think we just decided to go our own way though. We can always go back to it if there is enough enthusiasm for it. It was a nice easy way to allocate IPs based on location, and there was easily enough for a few devices in each street IIRC.
Cheers,
Andy
the 10.21.*.* schema we are using looks suspiciously like the consume range, was that just a case of mistransposition (if there is such a word!)?
Good point – you may be right there. Perhaps a typo back when it was done? Alternatively, it could be coincidence 😉
Cheers,
Andy
dunno why i picked 10.20 10.21 10.22 and so on, random really
looking at that diagram again, it is a bit confusing that the colours of the red and green side of the ipcop are reversed, also I though Red was on the greenside of the ipcop, is it not? Correct me if I’m wrong.