[MLB-WIRELESS] ip allocation list based on suburb

[Simon J Mudd]

lotto at impulse.net.au (Will Lotto) writes:

[snip]

> There are obviously too many suburbs, and too many postcodes for easy
> allocation.
> We could subnet the 10. network so you have space for 660 address in
> the network (10 bit network field, instead of the 8 to get 255)... or
> use a wasteful solution like 10.30.68.xxx (for postcode 3068) ... OR,
> councils can be used.

Hey guys, why don't you assign subnets on an as-needed basis?  Is it
that difficult?  That way you don't need to pre-allocate the whole
10.x.x.x network, but just a tiny fraction of it.

I can see some sense in perhaps allocating multiple blocks of
addresses for an area (8, 16 or 32?), so rather than than allocating
what I think you've defined as the standard block size of 16
addresses, assign 32x16 addresses and then for people in the same area
assign them the next free standard block which is not used from this
range.  When you use up all 32 blocks go back and request another
32x16 block.

I think most people want to do things this way because they think
"static", ie they think of the complexity of setting up static
routing.  If you really are thinking about having a large city-wide or
nationwide network static links just don't work.  Those who are not
connected to anyone else can use whatever addressing they want and
those who are only need to use a routing daemon, maybe something as
simple initially as routed to perform RIP.  RIP isn't too hot when the
network grows and has several limitations, but it gets you going.
Ideally something like OSPF scales much better and when the network
changes adjusts to the new structure more quickly.

This software isn't that difficult to setup.  Someone probably needs
to write a HOWTO, but doing things this way means that you don't need
to think IP addresses really except for their allocation.  Who knows
Microsoft's or Yahoo's web servers' ip addresses (without cheating)?
No-one.  What you do is connect to the machine by using www.yahoo.com.
The same can be done with your network.  This _does_ mean that you
need the DNS, but it's not difficult to assign "standard names" to the
ips that are allocated to a node, which the addresses are requested,
and from then on assuming the routing is setup correctly you can
forget the ips and everything just works like it does on Internet.
These guys weren't stupid and their network has scaled remarkably
well!

I'm surprised you guys just can't see this.

P.S. I've written a version of the DNS robot used on the ampr.org
domain which could be setup to automatically assign names to each
active node and the associated client ip addresses.  If anyone is
interested in the details let me know.

P.P.S. Why is everyone complaining about the speed of this list?  If
there are only 250 subscribers and 40 messages a day whichever machine
is providing the service must have a very limited bandwidth.  All the
mesages are small so I can't really see what the problem is.  Maybe
it's the mailing list software isn't up to it?

Simon
--
Simon J Mudd,   Tel: +34-91-408 4878,  Mobile: +34-605-085 219
Madrid, Spain.  email: sjmudd at pobox.com,  Postfix RPM Packager

--
To unsubscribe, send mail to minordomo at wireless.org.au with a subject of 'unsubscribe melbwireless'  
Archive at: http://www.wireless.org.au/cgi-bin/minorweb.pl?A=LIST&L=melbwireless
IRC at: au.austnet.org #melb-wireless