A. General information
1. Features
The design of a WAN involves a number of specific constraints that fundamentally distinguish it from a local area network:
· When choosing a WAN service, it is important to take into account network latency and cost considerations. WAN services are generally more expensive than LAN technologies, largely due to the distance involved and the need to engage telecommunications providers.
· As various operators may be involved, it is essential to be familiar with communication standards and to adhere to standardised interfaces.
· We need comprehensive addressing schemes covering several thousand units, including IP addresses and the associated routing protocols.
· An additional security measure may be required to restrict access to certain users, particularly via VPNs (Virtual Private Network) or encrypted connections.
2. Services WAN
WAN services are responsible for transporting data over long distances, with the assistance of a network provider. These services can be billed in various ways, but generally there are three main types:
Fixed services (analogue or digital leased lines) are dedicated private circuits between two points. They are billed on a fixed rental basis, regardless of whether the connection is used or not. This type of service offers guaranteed bandwidth and consistent latency, making it particularly suitable for mission-critical applications.
Switched services such as standard telephone lines or ISDN lines establish a connection between two end points on demand. In this case, a connection charge is added to the line rental fee whenever a call is made. If no data is transmitted over the line whilst a call is in progress, the charge still applies. This type of service is now rarely used.
Packet-switched services carry user data to its destination exactly as and when it is presented to the network. The user pays for the actual data transmitted. This method enables a more efficient use of network resources through statistical multiplexing.
In switched and packet-switched WAN environments, the customer may or may not pay directly for the required bandwidth. Costs can therefore be reduced by removing redundant data through compression.
3. Operating modes
Regardless of a network’s physical architecture, there are two different operating modes: connection-oriented mode and connectionless mode.
Connection-oriented mode: All communication between two devices follows a three-phase process. The sender requests the establishment of a connection by sending a special data block. If the receiver accepts, a connection is established by setting up a virtual circuit in the network linking the sender to the receiver. The data is then transferred from one end to the other, after which the connection is released. This mode guarantees a dedicated path and the order in which data arrives.
Connectionless mode: data packets, known as datagrams, are transmitted without first checking whether the destination device, and any intermediate nodes, are active. It is then up to the network management equipment to route the message in stages and, if necessary, delay it until the recipient is active. This mode offers greater flexibility and resilience, but with no guarantee of delivery or order.
4. Modern WAN architectures
Today, enterprise wide-area networks rely on a combination of technologies provided by operators (MPLS, dedicated fibre, Internet links) and overlay mechanisms (overlay) such as IPsec or SSL VPNs. The aim is to securely interconnect remote sites, data centres and cloud resources, whilst optimising costs and service quality.
Legacy architectures based solely on costly MPLS links are gradually being replaced by hybrid solutions combining multiple access methods: high-speed fibre, 4G/5G and legacy xDSL. WAN routers can automatically switch traffic from one link to another in the event of a failure or congestion, according to defined policies (prioritisation of critical applications, path optimisation).
SD-WAN (Software-Defined WAN) solutions provide a centralised control layer that enables dynamic management of routing paths between sites, automatically encrypt traffic between branches and measure performance in real time (latency, jitter, loss). They are particularly suited to enterprises spread across numerous sites and to the massive use of high-performance applications. across multiple sites and to the wide spread use of cloud-hosted applications (SaaS, IaaS).