The enterprise wireless market started with fat or autonomous APs. These were perfect for small scale wireless networks with limited clients per access point. Indeed, each fat AP needed to be individually configured for radio parameters, access services, security and control. Clearly, that approach was not suitable for deployments in mid to large enterprises or organizations. It became obvious that some type of centralized AP control was needed and the WLAN controller was developed.
The WLAN controller (a rack mounted appliance or server) works with thin APs that retrieve their firmware and configuration from the WLAN controller which offers a single point of management for the entire wireless network. The WLAN controller also acts as a switch and as a firewall for all the wireless traffic that is tunneled through the controller. This offers single point of control and termination for all wireless traffic. WLAN controllers require powerful appliances so they can also perform advanced functions such as radio parameters auto-configuration (channel selection, power emission, etc.), intrusion detection and prevention, spectrum monitoring and analysis.
That being said, the “controller” option has never been an end in itself. It was the only means at that time, in the early 2000s, to convince the largest customers to embrace the WLAN technology and deploy WLAN networks that were able to meet the challenges they faced around centralized control, security, and resiliency. At that time, indeed, the limitations and costs of physical components in the APs could only lead to a centralized approach. Since then, chipsets, memories and processors have evolved to become more powerful and economical. Today, it is possible to virtualize the controller, deploy it and run it in a distributed way in the APs themselves with coordinated intelligence. This is the choice Alcatel-Lucent Enterprise (ALE) has made: Smart and advanced APs are managed as a single system or cluster and handle the control and forwarding planes in a distributed and coordinated manner.
The Unified Management and Distributed Control solution allows you to do all the functions of a centralized controller. In addition, it eliminates architecture complexity, single points of failure, traffic bottlenecks, latency and high operational costs. Eliminating the previously required controller from wireless deployment architectures offers many potential benefits to organizations and their IT departments.
Controller-based architectures involve high upfront capital expenses. They also involve high licensing and maintenance costs. The most obvious benefit of the distributed control architecture is the CAPEX that is reduced since no controller is required. The saving is even more significant for deployments that involve multiple controllers for redundancy or load-sharing purposes.
Additionally, the Alcatel-Lucent Enterprise licensing model foresees one single license per AP for management. This single license includes all features required today for a state of the art wireless network (intrusion detection, firewalling, deep packet inspection…), thus reducing software costs. It also brings simplicity and clarity in comparison with traditional licensing models that come with controllers and charge licensing fees per feature.
Having no controllers means less equipment to operate and manage providing several OPEX benefits: less rack space, less power and cooling requirements, no maintenance fees (especially for unused backup controllers), and, basically, less equipment to be monitored by the IT department.
In a centralized controller-based architecture, the controller is a single point of failure for the entire wireless network impacting all wireless traffic when the controller fails. The only way to minimize the impact is to add additional redundant controllers, but this comes with a high cost. With a distributed control architecture, that single point of failure does not exist. Indeed, the controller function is no longer centralized but shared by all APs in the management domain. When an AP fails, the neighboring AP will detect that and react by increasing their transmit power, thus avoiding any hole in the radio coverage. The impact will be purely local: Only clients associated with the failed AP will associate to the neighboring AP and authenticate again.
No traffic bottleneck and decreased latency
A WLAN network is now a critical and indispensable asset for an organization. Wi-Fi is no longer a comfort option only. The WLAN is expected to connect bandwidth-hungry and/or latency sensitive applications, such as voice or video over IP, and video streaming). Over the years, the technology has improved to provide increasing levels of throughput with IEEE standards 802.11a/b/g/n and now 802.11ac that provides more than one Gigabit throughput over the air. In order to fully exploit the capabilities of 802.11ac APs,each AP should be connected to the LAN with IEEE 802.3bz 2.5GBase-T link providing up to 2.5G connectivity. Tunnelling such amount of traffic from each AP towards the controller will be difficult to sustain and may create a throughput bottleneck in addition to added latency. With distributed control approach, the traffic is no longer tunneled to centralized equipment but directly bridged into the local Ethernet switch.
When the maximum number of APs that a controller can manage is reached, deploying additional APs requires an additional controller. The distributed control architecture offers much better scalability: No controller equipement is needed, regardless of the size of the deployment.
Last but not least, the distributed control architecture is certainly the shortest route to the next breakthrough in enterprise wireless technology: Cloud Wi-Fi.
Do you think this is just a dream? Make it come true with Alcatel-Lucent Enterprise OmniAccess Stellar!
Learn more at www.al-enterprise.com/en/stellar