Compare traditional networks with controller-based networking

Traditional (distributed) networking

In a traditional network every router and switch is an independent, self-contained brain. Each one runs its own routing protocol process, its own spanning-tree process, and its own ARP cache. Each one makes its own decisions about how to forward traffic based only on the information it has learned from its neighbors. This is called a distributed architecture because the intelligence (the control plane, defined in section 6.3) is spread across every device.

Management in a traditional network is also distributed: the administrator configures each device individually through its CLI (console, Telnet, or SSH), or perhaps through a per-device web GUI. This is often called box-by-box management. There is no single place that knows the whole network’s configuration or intent; the “network configuration” is just the sum of 300 separate device configurations.

Advantages of the traditional approach include maturity (these protocols have been refined for decades), no single point of failure for the control plane (if one router dies, the others keep routing), and no dependency on a controller product. Disadvantages include slow, manual, error-prone changes, configuration drift, and difficulty seeing or enforcing a network-wide policy.

Controller-based networking

In a controller-based network, a central piece of software called a controller holds the network-wide view and policy. The administrator (or an application) tells the controller what the network should do; the controller then communicates with each device to make it happen. The devices may still run their own protocols for actual forwarding, but the policy and management are centralized.

Cisco’s enterprise controller is Cisco Catalyst Center (renamed in 2023; it was previously called Cisco DNA Center). The exam may use either name. Catalyst Center is the controller for Cisco SD-Access (software-defined access, for campus LANs) and provides automation, assurance (monitoring and analytics), and policy for the campus. For the WAN, Cisco’s controller-based solution is Cisco SD-WAN (from the Viptela acquisition), where the vManage/Catalyst SD-WAN Manager controller centrally manages WAN edge routers. In the data center, Cisco ACI (Application Centric Infrastructure) uses the APIC controller.

Side-by-side comparison

Why the exam cares

Cisco’s direction for enterprise networks is controller-based. The exam tests whether you understand the trade-off: centralized policy and automation in exchange for a dependency on a controller. Questions typically ask you to identify a characteristic of one model versus the other, or to name the Cisco controller for a given environment.