8 Network Architectures
Understand how network devices and services are organized, compare service-sharing models, and trace how common network designs deliver traffic.
What describes
A describes how network devices, services, and connections are organized to provide communication. It includes both the roles of participating computers and the design of the network that connects them.
These are related but distinct concerns. Client-server and peer-to-peer describe how devices share services, while designs such as hierarchical campus and describe how network equipment is arranged.
How devices share services
In the , a client requests a service and a server provides it. A server is a computer or system performing a service role; it may be a dedicated machine, a virtual machine, or a cloud service. For example, a laptop can use to look up a website's IP address and then request a web page from a web server. In a business, clients might request shared files, printing, or applications from centrally managed servers.
Centralized services make administration, access control, backups, and updates easier to coordinate. However, service depends on the server and its network connection being available, so organizations may use redundancy and backups. A network can have many servers, each providing a different service.
In a , devices communicate and share resources directly; each device can act as both a requester and a provider. For example, two computers on a small home network might share files without a dedicated file server. P2P can be simple and inexpensive for small groups, but managing permissions, backups, and availability becomes harder as the network grows. A peer-to-peer network may still use centralized services such as a router or server; the term describes how the shared resource or application works, not necessarily every service on the network.
The models differ in where services are provided and how administration is organized:
Service location: Client-server services are usually provided by designated servers; peer-to-peer resources are shared directly among participating devices.
Administration: Client-server administration is often centralized, while peer-to-peer administration is often distributed among peers.
Common fit: Client-server is common for organizations and services used by many clients; peer-to-peer is common for small groups or applications designed for direct sharing.
Main design concern: Client-server designs must consider server capacity, availability, and security. Peer-to-peer designs must consider peer availability, coordination, and consistent protection.
Common network designs
A typical home or small-office network connects computers, phones, printers, and other devices over Wi-Fi or Ethernet to an access point or switch. A router connects the local network to the internet and forwards traffic between networks. The router may also provide services such as address assignment through . Devices may share directly with one another while also using client-server services such as websites, cloud applications, or a network printer.
Hierarchical campus networks
Schools and businesses often organize a campus LAN into layers:
: Connects end devices such as computers, phones, and wireless access points.
: Aggregates access networks and commonly applies routing, policies, and other controls.
: Provides fast, resilient transport between distribution blocks and other parts of the network.
Large campuses may use all three layers. A smaller site can combine the distribution and core functions into a . This layered approach makes expansion and fault isolation easier: a problem in one access block need not require redesigning the whole network.
Data-center networks
Data centers must connect many servers and support substantial server-to-server traffic. In a design, servers and other endpoints connect to leaf switches; each leaf connects to the spine switches, which interconnect the leaves. This creates multiple paths through the fabric and supports predictable connectivity as the data center grows. Routers or other edge devices connect the fabric to external networks.
Branch and wide-area networks
A company with multiple offices connects each office's local network to other sites and shared services over a . At a branch, endpoint devices connect through access switches or wireless access points; a router or gateway forwards traffic between the local network and the WAN or internet. Firewalls and other security functions may control traffic at the boundary. A central site or cloud service can host applications used by branch clients.
How network components work together
Consider an employee opening a company application hosted in a data center. The laptop connects by Wi-Fi to an access point, which carries traffic through a switch in the campus . A router forwards traffic toward the data center or WAN. Switches move traffic within local network segments, while routers forward packets between IP networks. The data-center network carries the request to the server hosting the application, and the server sends a response back along a routed path.
Services such as can provide a client with network configuration, while can translate a service name into an IP address. These services support the application exchange but are separate roles from the switches and routers that carry traffic. In real designs, firewalls, redundant links, backup servers, and wireless systems may be added to meet security, availability, and capacity needs.
Putting the designs in context
Client-server and peer-to-peer models describe how devices provide and use services; network architectures describe how those devices and services are connected. Small networks commonly combine direct sharing with centralized services. Campuses use layered designs, data centers commonly use fabrics, and branch networks connect local devices to remote services over a WAN. In every design, endpoints, switches, routers, servers, and supporting services work together to deliver traffic to its destination.