Subnet use cases and considerations for network performance

Subnet use cases to improve network performance

Network administrators primarily create subnets to optimize network performance, but additional reasons for subnet use exist beyond performance. Other subnet use cases include the following:

• Network traffic management.
• Branch office organization.
• Cloud optimization.
• Security and privacy.

Network traffic management

Network administrators might create subnets to help manage traffic flow. When a device sends data across a network, the data is encapsulated into packets. Each packet includes a header that includes a source and destination address, among other things.

In a legacy network, without intelligent switches, packets traverse an entire network. Each host on the network examines the packet's header to see if it's the intended recipient. If the recipient address belongs to a different host, the unintended hosts ignore the packet. If a host is the intended recipient, it opens the packet and processes the data within.

If two hosts are in close proximity and frequently communicate with one another, it might not make sense for the two hosts to send information that traverses the entire network. Instead, it might make sense to place the two hosts in a dedicated subnet that can isolate traffic flow between the hosts to that particular subnet and relieve congestion on the rest of the network. The two hosts can still communicate with other machines outside of the subnet, and traffic crosses subnet boundaries when it is specifically destined for a host in a different subnet.

Branch office organization

Subnets mimic the physical design of a network, which helps enterprises organize branch office networks. If an organization has branch offices connected to the corporate office via a leased line or a point-to-point VPN, the organization can create separate subnets for each branch office.

If a branch office and the corporate headquarters are part of the same subnet -- a design known as a flat network -- the link between the two facilities could become congested with unnecessary traffic. If an organization segments the branch office in a separate subnet, however, it can improve network performance and enable organizations to organize their resources.

Cloud optimization

Enterprises also commonly apply subnets to cloud resources. If an organization provisions resources in a public cloud, those resources can reside in a subnet separate from the resources in the organization's data center. The use of subnets in the cloud also enables network administrators to group cloud resources, which gives them greater control of their cloud infrastructure and helps them manage and monitor resources more efficiently.

Security and privacy

Enterprises also commonly use subnets to ensure security or privacy. For example, an organization can create a dedicated subnet for server replication traffic. Replication traffic is sensitive, so network administrators typically isolate it to a dedicated subnet to prevent it from entering the organization's user network. These types of subnets typically aren't physically connected to the organization's main network, which prevents access to outsiders.

Organizations also sometimes create dedicated management subnets through which network admins can access back-end systems without exposing those systems to the user network.

Plan subnet placement

Once network administrators evaluate subnet use cases, the next step is to consider how to configure them in a network. Network administrators should take time to figure out how to subnet networks before configuration. This requires an assessment of business considerations that help determine placement.

Subnet placement considerations include the following:

• Geographic structure.
• User location.
• Hardware placement.
• Server placement.

Geographic structure

The structure of a subnet should mimic the network's geographic structure. Any cloud networks or facilities separated by WAN links should be in separate subnets. This isn't a necessity, though. It's possible to have one subnet that spans multiple facilities. In most cases, however, networks typically perform better if each facility uses its own subnet.

User location

The concentration of users in a department is another consideration for subnet placement. For example, it might not be clear where network administrators should place subnets in environments with thousands of users.

Some companies create subnets by department, but this isn't always ideal because departments often move to other parts of buildings. It might be better to base subnets on characteristics that are unlikely to change, like by floor or area.

Hardware placement

Network administrators also need to think about hardware placement before they create subnets. Every subnet should be linked together through routers, so network administrators must figure out how they plan to place routers before configuration.

Router placement considerations include the following:

• Size based on the anticipated traffic volume.
• Cost.
• Placement.
• Location of the connection to each subnet.

Server placement

It typically takes users longer to access servers in different subnets. Latency usually isn't enough to be a problem unless resources are geographically separated, such as if they're located in different clouds or buildings. However, if a network has hundreds of users who cross subnets to access servers, the router could become a bottleneck and create latency.

Network administrators have a few options to fix this problem:

• Servers restricted to one department. Place them in a subnet with the end users who access them most frequently.
• Servers accessed across the entire organization. Put multiple network interface cards into each server and bind each NIC to a separate subnet. It's also important to purchase routers with sufficient throughput to handle the traffic.
• On-premises networks. Give each subnet its own domain controller. Be sure to define sites through Active Directory to require users to authenticate through a domain controller in their own subnets or sites.

Are subnets still relevant in modern, switch-based networking?

Subnets offer many advantages, such as helping to relieve network congestion. But modern networking has largely eliminated network congestion. In legacy, hub-based networks, nodes connect through a hub, which create collision domains. Excessive collisions can slow down a network. Network switches fix this problem because they direct packets from the source to the destination machine.

Although switches help fix the problem of collisions -- and, in turn, largely eliminate network congestion -- subnets still play an important role in modern networking. Subnets remain relevant in switch-based networks and can help in ways switches can't.

For example, switches help performance when a packet is destined for a specific device, but every device on the network still receives broadcast traffic. Subnets can help organize a network with many broadcast networks. Network administrators can separate them based on department, location, resource allocation or other criteria. The extent to which an enterprise chooses to use subnets depends on the network design and its performance.

In short, subnets aren't obsolete and remain widely used for several reasons. Subnets can support route summarization, which helps optimize network traffic and improve overall network efficiency. Subnets can also isolate certain types of network problems, which makes it easier for network administrators to troubleshoot.

Brien Posey is a 22-time Microsoft MVP and a commercial astronaut candidate. In his over 30 years in IT, he has served as a lead network engineer for the U.S. Department of Defense and as a network administrator for some of the largest insurance companies in America.