In one of our last segments, we discussed the various WLAN topologies in common use. This segment will discuss the concept of WLAN architectures. While a topology essentially describes the physical configuration of a WLAN, or any LAN for that matter, architecture describes a design concept within which a topology can exist. Network architecture describes the logical relationship between network entities, while a topology describes the actual physical connection required to achieve the logical design. WLANs are described by three broad categories of architectures: autonomous, centralized (controller-based), and cooperative (controller-less).
In an autonomous architecture, access points (APs) are stand-alone (sometimes called "fat") APs that contain all the necessary features and capabilities to operate without any reliance on another device. An autonomous AP operates on all three network planes: management, control, and data. Autonomous architecture allows for several APs to connect to the wired infrastructure and provide a portal for its basic service set (BSA). Multiple autonomous APs can be interfaced to the same infrastructure and form an extended service set (ESS). If all autonomous APs in an ESS adhere to the 802.11 standard, then the APs will cooperate and communicate over the wired backhaul and provide seamless roaming of a client device without any active management. This architecture is suitable for small and small office, home office (SOHO) implementations properly configured to avoid adjacent and co-channel interference. Loss of any AP, however, creates a dead spot in the ESS that would not readily be recovered by nearby APs. Losing connectivity will affect adjacent APs and result in disassociated clients re-associating to an available AP. This will cause a rise in the client/AP ratio and a sudden increase in traffic.
The second type of architecture relies on a centralized controller to regulate the operation of the WLAN. The controller usually takes the form of a hardware device that either is wired to the APs at the network edge, or uses a wireless system to provide local connection to clients on one frequency, while performing control on another. Controller-based APs are referred to as "lightweight" APs and usually operate completely on the data plane. Data forwarding and routing, along with network configuration and management, are done by the controller, which operates on the control and management planes. The AP has diminished functions and is essentially just a radio transceiver in the production area. A lightweight AP communicates with the client devices or other APs, but under the strict authority of the controller. Controller-based WLAN architecture requires a full-time, dedicated controller for the network to function.
The vast majority of wireless vendors use a wireless network management system to control wireless APs on a large network. Small networks typically don’t need this level of management, which includes advanced features like customized security, captive portals, and VLANs. Large networks servicing hundreds or thousands of clients require a means of controlling the network operation to ensure reliability, security, and quality. All large enterprise-level networks use management systems. In contrast to the autonomous architecture described above, a centrally controlled network has the ability to dynamically set channel assignments, AP power output, and client load balancing. In the event that an AP is out of service, the controller will increase the power output of adjacent APs to provide coverage, while at the same time balance the client load for optimum performance.
In some cases, the functions of the data plane are split between the AP and the controller, resulting in what is called "split-MAC" architecture. This simply means that some of the data processing function, such as access control, is done at the controller level and not the AP. In split-MAC, the AP has the capability to perform basic access and switching functions, but still needs to rely on a controller for higher level network functionality, such as authentication.
Controller-based systems, however, suffer from a simple weakness in that the controller device presents a single point of failure; mitigation requires an additional, redundant controller, preferably in another location. A redundant controller requires additional capital expenditure, management, software licensing, and overhead. However, in a secure facility with a well-designed and installed network infrastructure, controller-based systems provide excellent performance, robust security, and feature-rich management.
Controller-less, or cooperative, architecture is based on the use of virtual management (cloud-based) systems that utilize a minimum of wired APs and relies on a cooperative communication method between APs to manage and control a WLAN. These systems rely on cooperative routing and messaging protocols to provide control of and between full-featured APs. APs on these systems occupy both the data and control planes; only the control software exists in the management plane. As a result, the controller is virtual and passive; control and data are handled by the APs. If the controller is put out of service, the network will continue to operate. As in a controller-based WLAN, the loss of an AP will be addressed by an increase in power form adjacent APs. There is no controller, so the problem of network failure from the loss of a single device is not present. The management interface provides full-featured configuration and control of the APs and the network from anywhere; the system administrator does not have to be on premises to access and control the network.
Typically, a mesh topology will implement a cooperative architecture. The idea of a self-managing, forming, and healing WLAN is ideally suited to this type of implementation. The combination of architecture and topology, in this case, results in a very robust network, and one that is very well suited to the rigors and requirements of distributed control systems (DCS). Because the requirements for wired backhaul are minimal, APs can be placed in areas that would ordinarily require expensive wired interfaces. However, an over reliance on the mesh topology can result in a sudden attenuation over several "hops" or repeats from AP to AP. Care must be taken to avoid more than three hops to avoid compromising performance.
There is much debate on the pros and cons of the two latter architectures. The difference between the two types is vast: there are cost, operation, and reliability issues that are unique to each. Each method has its pros and cons, and each has its partisans.
– Daniel E. Capano, owner and president, Diversified Technical Services Inc. of Stamford, Conn., is a certified wireless network administrator (CWNA). Edited by Chris Vavra, production editor, CFE Media, Control Engineering, firstname.lastname@example.org.
www.controleng.com/blogs has other wireless tutorials from Capano on the following topics:
The management, control, and data planes
www.controleng.com/webcasts has wireless webcasts, some for PDH credit.
Control Engineering has a wireless page.