Technology Blog

The Evolution of WLAN Architecture


A recent visit to the Intel Tech Museum in Santa Clara got me thinking about WLAN architectures. OK, I admit…I think about WLAN just about everywhere I go, but the Tech Museum gave me an apt analogy.


The history of WLAN architectures draws close parallels to the history of computing. Computing began with an electrically driven mechanical calculator, the Z1, in 1938. Then scientists realized the power of pooling and centralizing computing resources. That, combined with advances in semiconductor technology ushered in the age of mainframes. And then Moore’s law combined with the need to un-tether engineers from the mainframe brought along the era of distributed computing which we now live in. In this evolution, which has spanned decades, it is now established that both the centralized model and distributed models retain their relevance. Tablet computers are cool and convenient, but an iPad can never solve problems in quantum physics, climate research or space exploration or for that matter, beat Gary Kasparov at a game of chess. Each model is irreplaceable in its niche.


Ditto with wireless LAN. The early access points were disjointed and uncoordinated. While they provided basic access, they offered low throughput and were a management and security nightmare. Then came the controller-based WLAN solutions that reduced the role of access points to mere radios. For the first time, WLAN offered performance, scalability, manageability, security and most importantly, mobility! As the CPU on the APs scaled from MHz to GHz and RF technology scaled from 11 Mbps to 450 Mbps (and beyond, soon) it is inevitable that the AP will be able to shoulder more of the functions traditionally performed by controllers. But that does not mean controllers will be artifacts of history. The overlay model that made centralized architectures immune to messy underlying wired networks will continue to be a huge driver for controller-based solutions. There are many more reasons in favor and against both architectures that I will address in a future blog.


As somebody who lived through the evolution from legacy to 802.11n, experienced the controller world and who now manages a controller-less product line, I am equally excited for the future of both architectures. I am just glad to be at a company which embraces both architectures and recommends to each customer what is best for them. At Aruba, we try not to re-shape the foot to fit the shoe ☺



Occasional Contributor I

I too was around in the early days of WLAN for a 2500 AP deployment on the Microsoft campus in early 2000. 802.11b had just been ratified in September 1999 and only 2 802.11b products were available – Lucent and Aironet. We chose Aironet because of their strong SNMP support. Our first building with 60 access points took 8 engineers 16 hours to configure and verify. We developed software and NMS tools to simplify the configuration and maintenance and bolted together various components to manage that many APs.


Late in 2000, at an IEEE meeting in Tampa, an engineer from Cisco approached me to talk about my deployment and how an enterprise level WLAN system could address the pain points. We chatted about automatic frequency assignments and power levels, reducing the number of managed devices and overall wireless security. In 2001 Blackstorm Networks (became Airespace) was formed by a couple of very active IEEE members. In 2002, Merv, the Cisco engineer I spoke about earlier, told me that Cisco had rejected the controller based architecture and that he and some others had formed Aruba Networks. The rest is history.


I couldn't agree more about the different architectures, but I would add that there is likely to be a continuum between the central controller and the controllerless environments. As WLANs start to scale far beyond “thousands of APs” to millions of APs, a hybrid system that includes control (MCs), Monitoring (AirWave), Authentication (ClearPass) and Scale (Instant) will be necessary. In the not too distant future extremely large deployments are likely with multiple access points in every home and business. Electric  companies could include APs as part of a “Smart Grid” and other network providers are already incorporating APs into their products. The top 10 electric companies have over 32M customers.


The future is bright for those that think big.





Your post brings out one more highlight of this evolution - back in the day, it took an electrical engineer to operate a the e-trade baby uses an iPad. If WLAN were to follow in those footsteps, it would not require a network engineer to set up and operate an enterprise WLAN! Subject of my next blog post :)

Search Airheads
Showing results for 
Search instead for 
Did you mean: 
Read all about it! If it’s happening now, it’s in the community.

Check out the latest blogs from your community team, the community experts and other industry sources.