Why SDN Became the Hottest Trend: From OpenFlow to Overlay Networks Explained

Problems with Traditional TCP/IP Networks

Traditional networks suffer from long innovation cycles, complex and numerous protocols, and weak path‑planning capabilities. The control plane and forwarding plane are tightly coupled, making large‑scale updates slow and costly, while the proliferation of standards such as OSPF and BGP increases operational difficulty and expense.

What Is SDN?

Software‑Defined Networking (SDN) separates the control plane from the forwarding plane. A centralized SDN controller, implemented in software, acts as the network’s brain, issuing forwarding decisions to simple forwarding devices (switches/routers) that no longer need their own control logic.

OpenFlow: The First SDN Protocol

OpenFlow was introduced by the Stanford Clean Slate project (2006) and commercialized in 2009. It defines a secure TCP channel between the controller and forwarding devices, uses TLS for encryption, and requires periodic hello messages for link health. The protocol standardizes how the controller collects device statistics (ports, bandwidth, link failures, utilization, table size) and installs flow entries that dictate how packets are handled.

Flow‑Table Evolution

Early OpenFlow 1.0 flow tables matched many fields, leading to exponential growth in entry count and high TCAM cost. For example, matching source and destination IPs for three sources and three destinations requires nine entries; adding a city field doubles the entries to eighteen. OpenFlow 1.1 introduced multiple tables, allowing separate matching stages (e.g., source then destination), reducing entry count from nine to six in the same scenario and improving lookup speed and storage efficiency. Subsequent versions (1.2‑1.4) added more match fields and support for multiple controllers to increase resilience.

Overlay Technologies and VXLAN

Overlay networks create a logical network independent of the underlying physical infrastructure. VXLAN encapsulates Layer‑2 frames inside UDP/IP packets, enabling virtual machines to reside on any physical host while appearing to be on the same broadcast domain. This approach avoids re‑engineering the physical network, sidesteps the limitations of spanning‑tree protocols, and supports large‑scale data‑center deployments.

Future Directions and Industry Debate

Two camps have emerged: the “revolutionary” side advocates full control‑plane/forwarding‑plane separation using OpenFlow, while the “improvement” side prefers overlay‑based solutions that enhance existing networks without drastic hardware changes. Both seek greater flexibility, but each faces challenges—OpenFlow’s incomplete flow models and security concerns, and overlay’s integration with legacy routing and gateway functions. The ongoing competition suggests that SDN will continue evolving, possibly blending both approaches.