Keith Shaw

Wide-Area Networks: What WANs Are and Where They’re Headed

If it weren’t for wide-area networks, it wouldn’t be possible to create unified networks for organizations with far-flung locations, to telecommute, or to do online anything. But WANs do exist and have for decades, constantly evolving to carry more and more traffic faster as demands increase and technology becomes more powerful.

What is a WAN?

A WAN is a network that uses various links—private lines, Multiprotocol Label Switching (MPLS), virtual private networks (VPNs), wireless (cellular), the Internet—to connect smaller metropolitan and campus networks in diverse locations into a single, distributed network. The sites they connect could be a few miles apart or halfway around the globe. In an enterprise, the purposes of a WAN could include connecting branch offices or even individual remote workers with headquarters or the data center, in order to share corporate resources and communications.

WAN architecture

Initially, WANs were built with meshed webs of private lines bought from telecommunications carriers, but WAN architectures have advanced to include packet-switched services such as frame relay, ATM and MPLS. With these services, a single connection to a site can be connected to many others via switching within service-provider networks. For certain types of traffic, the Internet can also be woven into the mix to provide less expensive WAN connections.

Software-defined or SD-WAN

As companies look for WAN improvements, the use of software-defined technology is gaining momentum. Software-defined WAN (SD-WAN) takes software-defined concepts, especially the decoupling of the control plane from the data plane, and brings it to the WAN.

SD-WAN uses software to monitor the performance of a mix of WAN connections—MPLS, dedicated circuits, the Internet—and to choose the most appropriate connection for each traffic type. So teleconferencing might run over a dedicated circuit, but email might use the Internet. In making its decisions, SD-WAN software takes into account how well each link is performing at the moment, the cost of each connection and the needs of each application.

Many believe that SD-WAN is poised to take off in 2018, moving from an early adopter technology to mainstream implementation. Research firm IDC has predicted that SD-WAN revenues will hit $2.3 billion in 2018, with a potential revenue target of $8 billion by 2021. The first phase of SD-WAN aimed at creating hybrid WANs and aggregating MPLS and Internet connections to lower costs; the next phase will improve management, monitoring and provide better security, according to Lee Doyle of Doyle Research.

A subset of SD-WAN called SD-Branch will help reduce the need for hardware within branch offices, replacing many physical devices with software running on off-the-shelf servers. Mobile backup across a SD-WAN can provide a failover for broadband connections as wireless WAN technology (4G, LTE, etc.) costs decrease.

WAN protocols

One of the earliest protocols used to deliver WAN traffic is X.25, which uses packet-switching exchange nodes (PSEs) for the hardware that drops traffic onto the wires connecting sites in standard-sized packets, delivered in order, and includes error correction. The physical links include leased lines, dialup telephone services or Integrated Services Digital Network (ISDN) connections. It’s not used much anymore.

Frame relay is a successor to X.25. Frame Relay places data into different-sized frames and leaves error correction and retransmission of missing packets up to the endpoints. These differences speed up the overall data rate. In addition, Frame Relay relies less on dedicated connections to create meshed networks, meaning fewer physical circuits, hence saving companies money. Again, frame relay, while once extremely popular, has become less so.

Asynchronous Transfer Mode (ATM) is similar to frame relay with one big difference being that data is broken into standard-sized packets called cells. Cells make it possible to blend different classes of traffic onto a single physical circuit and more readily guarantee qualities of service. The downside of ATM is that because it uses relatively small cells, the headers eat up a relatively large percentage of the contents of each cell. Therefore, ATM’s overall use of bandwidth is less efficient than that of frame relay. ATM has also fallen out of favor with business customers.

Today, multi-protocol label switching is used to carry much corporate data across WAN links. Within an MPLS network, brief header segments called labels allow MPLS routers to decide quickly where to forward packets and to treat them with the class of service indicated by the labels. This makes it possible to run different protocols within MPLS packets while giving different applications appropriate priority as traffic travels between sites.

Internet protocol (IP), which became more ubiquitous in the 1990s, is one protocol commonly carried within MPLS.

WAN management and optimization

Because data transmission is still reliant on the rules of physics, the greater the distance between two devices, the longer it will take for data to travel between them. The greater the distance, the greater the delay. Network congestion and dropped packets can also introduce performance problems.

Some of this can be addressed using WAN optimization, which makes data transmissions more efficient. This is important because WAN links can be expensive, so technologies have sprung up that reduce the amount of traffic crossing WAN links and ensure that it arrives efficiently. These optimization methods include abbreviating redundant data (known as deduplication), compression, and caching (putting frequently used data closer to the end user).

Traffic can be shaped, giving some applications (such as VoIP) a higher priority over other, less urgent traffic (such as email), which in turn helps improve the overall WAN performance. This can be formalized into quality of service settings that define classes of traffic by the priority each class receives relative to others, the type of WAN connection that each traffic type will travel, and the bandwidth that each receives.

History of WANs

WANs have been around since the early days of computing networks. The first examples of WANs included circuit-switched telephone lines, but advances in technologies now include wireless transmissions (your cell phone basically operates on a wireless wide-area network, or WWAN) and fiber-optic transmissions. Data can also be moved via leased lines, or even via satellite transmission.

As technologies changed, so did transmission rates. The early days of 2400 bps modems evolved to 40 Gbps and 100 Gbps systems today. These speed increases have allowed more devices to connect to networks, witnessed by the explosion of computers, phones, tablets and smaller Internet of Things devices.

In addition, speed improvements have allowed applications to utilize larger amounts of bandwidth that can travel across WANs at super-high speed. This has allowed enterprises to implement applications such as videoconferencing and large-file data backup. Nobody would have considered conducting a videoconference across a 28Kbps modem, but now workers can sit in a cubicle and participate in a global company meeting via video.

Many WAN links are supplied via carrier services in which customers’ traffic rides over facilities shared by other customers. Customers can also buy dedicated links that nail up circuits point-to-point and are dedicated to just one customer’s traffic. These are typically used for top-priority traffic or delay-sensitive applications that have high-bandwidth needs such as videoconferencing.

Connections between WAN sites may be protected by virtual private networking (VPN) technology that overlays security functions including authentication, encryption, confidentiality and non-repudiation.

Interplanetary Internet

WAN technologies aren’t just limited to Earth. NASA and other space agencies are working to create a reliable “Interplanetary Internet,” which aims to transmit test messages between the International Space Station and ground stations. The Disruption Tolerant Networking (DTN) program is the first step in providing an Internet-like structure for communications between space-based devices, including communicating between the Earth and Moon, or other planets. But until we can achieve faster-than-light technology, the network speeds would likely top out at the speed of light.


This article was written by Keith Shaw from NetworkWorld and was legally licensed through the NewsCred publisher network. Please direct all licensing questions to