Reasons to Divide a Network Into Subnets
By Finn McCuhil
Most business networks operate well using a single network segment for their entire operation. Using a single IP segment, along with Dynamic Host Configuration Protocol (DHCP), is the simplest way to set up a small network in a single location. When the network is larger, separated into two or more locations or has other special requirements, a network administrator needs to investigate alternate setups to expedite network traffic. Subnetting, or breaking a network into smaller, discrete segments is a logical choice to handle several special situations.
High traffic volume can slow throughput time on a single segment network. Because each incoming and outgoing data packet is forced to search the entire range of IP addresses to find its assigned destination, there is a high collision rate when large volumes of traffic are competing for space across every node on the network. Higher collision rates create lost packets, higher latency and reduce overall transmission rates. Breaking down the network into smaller segments reduces the number of nodes any single packet contacts in its search for the proper destination. If, for example, the computers in the marketing department don't need to communicate directly with the computers on the manufacturing floor, both can be placed on separate segments. The result is an increase in throughput for both departments.
Placing individual departments on individual subnets incidentally improves security as well. Because computers on one segment don't connect directly with other segments, there is a reduced risk of a computer on a manufacturing subnet connecting directly to confidential information on a computer hard drive in accounting. Some network administrators choose to place all wireless clients on a single subnet for this reason as well. Restricting network access to potential network intruders is another layer for security in situations where broadcast signals could be subject to snooping or attack.
Different communications protocols on the same network segment are a cause of high collision rates as well. On a network using protocols other than Ethernet, moving legacy communications to a separate segment will speed up overall network performance. Networks using older token ring protocols would be excellent candidates for segmentation.
One of the most common and compelling reasons to segment a network is when you are routing traffic for wide area network for a remote location(s). Setting up office communications for a single network in multiple locations requires subnets for effective operation. Minimizing unnecessary traffic between remote routers is a primary requirement for maintaining adequate transmission speed between locations across town or across the country.
Finn McCuhil is a freelance writer based in Northern Michigan. He worked as a reporter and columnist in South Florida before becoming fascinated with computers. After studying programming at University of South Florida, he spent more than 20 years heading up IT departments at three tier-one automotive suppliers. He now builds wooden boats in the north woods.