Illustrated TCP/IP by Matthew G. Naugle Wiley Computer Publishing, John Wiley & Sons, Inc. ISBN: 0471196568   Pub Date: 11/01/98

Chapter 56
Subnet Mask Decisions

Let’s say you are assigned one network number and you are using RIP version 1. Although we are introducing this concept here, it is covered in more detail later. It is provided here to give you an understanding that under certain conditions, certain decisions have to be made. The subnet mask must be the same throughout your network, unless you change network IDs. You must make a decision on how large the subnet mask should be. How many hosts per subnet will there be? What about expansion? These are issues you must consider when assigning a subnet mask. With RIPv1, it is a trade–off. OSPF and RIPv2 do not have this trade–off, but care must still be taken when assigning network masks to a network number. This is shown completely in the next section on advanced IP addressing concepts.

This restriction becomes readily noticeable when assigning an IP address to a serial line (two routers using a leased phone line to connect). There have been circumstances that some router vendors have come up with that allow for the no IP address assignment for a serial line. However, if the serial link needs an address assignment and you are not using RIP version 2 or OSPF, a whole subnet number is wasted on this point–to–point link. A serial link will consume a network number and associated host IDs. Therefore, a unique network number will be assigned and, instead of being able to use all available host IDs, it will be possible to use only two host IDs (there will be only two addressable points on that network).

The rest of the host IDs will be lost for that network number and will be assigned and used for that serial link; therefore they will not be able to be assigned to any other links. If you have a large site that will encompass many serial links and you do not have the ability to assign a large number of network numbers, use subnet addressing and the routing protocol of OSPF. OSPF supports variable–length subnet masks, which will collapse that serial link into two hosts within a network number; therefore, no host numbers are wasted on serial links. Variable–length subnet masks allow a single network number to use multiple masks (unlike RIP version 1, RIP version 2 allows VLSM). This allows more bits to be assigned back to the network, allowing a more efficient use of the address.

A few more things you need to consider: If the network station moves to a new network, does the IP address for that station change? Like the current telephone system, IP addresses must change when the network station is moved to a new network that employs a different network number. If the network station is moved on the same logical network, the IP address may remain the same. For example, if a network station is moved to a different part of the same subnet, the whole IP address may stay the same. If the network station is moved to a different subnet (different subnet number), the IP address of the network station must change.

This subject will be picked up again in the section “Advanced IP Addressing.”