TCP/IP

Having TCP/IP present on a network has become commonplace in the past couple of years. TCP/IP (Transmission Control Protocol/Internet Protocol) has gone from being used almost exclusively at universities and large corporations to becoming the protocol of choice in networks of any size. With more networks connecting to the Internet on a daily basis, planning for implementation of TCP/IP while allowing for growth into the future is increasingly important. This article is designed to provide insight for  you as you go through the steps of planning for TCP/IP use on your network.

 Public vs. private IP addresses

There are two types of IP addresses: public and private. Public addresses are assigned to you by the ISP (internet service provider) you're using to connect to the Internet. These addresses are doled out to ISPs by Internic, based on the particular ISP's demonstrated needs. Getting a full IP network range is almost impossible even for the largest of companies. Compound this with an unknown rollout date for Ipv6 (the next version of IP addressing with a larger address space range to work with), and suddenly using private IP addresses—with the protection they provide from network hacking—start to become a viable option. The following recognized private IP address ranges are specified by Internet Request for Comments (RFC) 1918:

• 10.0.0.0         –     10.255.255.255

• 172.16.0.0     –     172.31.255.255

• 192.168.0.0   –     192.168.255.255

One of the features that comes with using the private IP addresses as specified in RFC1918 is that the major router vendors and users agree that requests going to or from these addresses are discarded. When using private IP addresses, you'll use some type of proxy or Network Address Translation (NAT) server to convert the private IP address range(s) on your local network to a public IP address that can be routed.

 Subnetting

Subnetting is the process of taking one TCP/IP network address range and splitting it into two or more ranges that can be used on as many network cards. The "Subnet Expansion Table"  (courtesy of http://www.certnotes.com/default.htm ) shows the effects of subnetting IP addresses.

Subnet Expansion Table

Subnetting, although becoming more common, should be done with careful consideration. Because some of the addresses ranges created by subnetting will start with either all ones or all zeros, you must remember that the first and last network address ranges are unusable for routing purposes. One disadvantage to using a subnet mask other than the default for that IP class is that it adds a layer of complexity to troubleshooting when a workstation or other network device has a communication problem.

Normally  IP addresses should be as short as possible and with the simplest subnet mask to minimize errors caused by mistyping. In cases where multiple network segments were required at a given location, we used additional address ranges from the third octet part of the IP addressing scheme. Don't subnet unless you have to, and then try to plan for future use as much as possible to avoid problems with having to change the IP configuration at a later time.

There's no one right or wrong way to use the IP addresses for a given subnet. We prefer to use .1 of the fourth octet of the address range (that is, 10.10.1.1) to identify the default route to and from a particular segment. Depending on how many devices will be present on a particular subnet, we like to assign addresses to those devices that shouldn't need to change or can't change for some reason.

Devices such as network printers or communications gateways usually need to have a fixed address. If a DHCP address release isn't renewed in time and expires, thereby causing a new address to be assigned to the device, the other devices or services talking to the particular device won't understand how to change to the new address.