An IP address names a machine on a network, and a settlement numbers its own network from a private range because no registry is left to hand out public addresses. Give each settlement its own block, for example one /24 subnet with 254 usable addresses, reserve the low addresses for the router and the servers, and write the plan on paper before the first machine is added. The plan, not the numbers themselves, is what keeps a growing network from colliding with itself.
Part of Networking
Logical addressing: how machines are numbered, grouped into subnets and documented, and why a settlement uses private ranges.
A Settlement Makes Its Own Numbering Because There Is No One To Ask
Every machine on an IP network needs an address that no other machine on the same network has. Network Addressing explains the three kinds of address a machine carries: the hardware address burned into its card, which only works on one segment; the IP address, which can be routed between networks; and the port number, which picks the service. This section is about the IP address.
On the old internet, public addresses were handed out by a chain of organisations, which is part of the institutional machinery that cannot be assumed to exist. A settlement does not need a public address to talk to its own machines or to a neighbour it is wired to. It needs addresses that are unique across the networks it joins, which it can assign for itself from the private ranges (Private Ranges).
What You Need
- A chosen private block for each settlement (Private Ranges)
- The arithmetic of subnets (Subnet Masks) to size each block
- Paper and a pen — Network Addressing lists what an address plan must contain: ranges, subnet allocations, a record per machine, reserved ranges and the names
Method: Draw Up An Address Plan
Step 1: Give each settlement its own subnet The carrier Internet Infrastructure uses 10.1.0.0/24 for the first settlement, 10.2.0.0/24 for the second, and so on. Each /24 supports 254 devices, which it calls more than enough for most communities.
Step 2: Divide each subnet by purpose The carrier reserves .1 for the router, .2 to .10 for servers (name, web, mail, file), and .11 to .250 for ordinary machines handed out automatically (DHCP).
Step 3: Number the links between routers separately A link between two routers needs only two addresses, so give each its own tiny subnet (CIDR Notation).
Step 4: Write it down Record each machine’s address, hardware address, name, location and purpose, and keep two copies in different places (Network Addressing).
Worked Example: One Settlement’s Plan
Subnet 10.1.0.0/24. Addresses .0 to .255 are 256 in all; .0 is the network address and .255 the broadcast address, which leaves 256 - 2 = 254 usable.
The carrier’s split: router 1 address (.1); servers .2 to .10, which is 10 - 2 + 1 = 9 addresses; ordinary machines .11 to .250, which is 250 - 11 + 1 = 240 addresses. Together 1 + 9 + 240 = 250. That leaves 254 - 250 = 4 usable addresses (.251 to .254) unassigned for later use.
Three settlements on 10.1.0.0/24, 10.2.0.0/24 and 10.3.0.0/24 can hold 3 x 254 = 762 devices between them.
Common Mistakes
- Two settlements using the same block — salvaged home routers often default to the same private network, and two sites that both use it cannot be joined without renumbering one; decide the plan first
- Two machines with the same address — a duplicate causes erratic behaviour; the plan prevents it
- An address plan that lives only on one disk — keep a paper copy in the equipment room, as the carrier says
What’s Next
- Learn next: IPv4 Structure — what the 32 bits of an address mean
- Combine with: DHCP — handing out the ordinary-machine addresses automatically