Applying the SDN principles to different parts of the network yields different subsets of SDN (i.e. SD-DC, SD-WAN…). The SDN principles are:

  • Direct programmability: you can interact with the network through well defined APIs
  • Agility: network topology changes are handled automatically and do not affect the system
  • Central management: pretty self descriptive. The management system should provide a global view of the whole network through a single interface. Cloud seems to be the technology of choice here.
  • Programmatic configurability: touching the device CLI is a big no-no. All changes to device configuration should be done by invoking API calls to the abstraction layer (either the management system or controller)
  • Vendor neutral: someday. Today, the best that we can get is multi-vendor support.

The question is — why apply these principles to the LAN? The LAN, although a comprehensive part of the network, is considered pretty static in terms of configuration changes and also predictable in behaviour. Switches gonna switch type of thing, right? It takes a technology that is very sensitive to end-to-end application connectivity, like Voice over IP, to show that even LANs are complex and often unpredictable in their behaviour. It gets even more complicated once you include Wireless LAN and consider the complex and dynamic nature of the RF medium. RF is never simple. Anyone that tried deploying Voice-over-WLAN will appreciate the intricacies behind how the RF medium affects end-to-end application connectivity. They will also recognise the amount of effort required to maintin the Voice-over-WLAN deployment during its lifecycle.

Once both wired and wireless LANs are considered and end-to-end application behaviour becomes important, the LAN becomes anything but static and straightforward.

The SD-LAN takes the SDN principles discussed above and applies them to the wired and wireless LAN to manage the underlying complexity. The goal is to provide faster rollout of network services, reduce downtimes, simplify operation and support interaction between the network services and user facing applications. Not unlike SD-WAN, SD-LAN also uses principles like dynamic path management to overcome changes in topology and even overlay tunnelling, something more commonly seen in SD-DC deployments, to extend the network when and where required.

SD-LAN is a pretty recent addition to the SDx landscape and I am sure it will evolve further. Some of the principles behind it, like agility, central management, overlay tunnels and even dynamic path management are not new and have been provided by WiFi vendors for years. However, they have been focused on the wireless part, often leaving out the wired network altogether. SD-LAN includes both, introduces a network abstraction layer and exposes a single interface for both management and programmability. Let’s wait and see what kind of improvements we see on the multivendor/vendor neutral front as well.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.