Bluesocket's vWLAN architecture continues the evolution of (and debate surrounding) the architecture of enterprise-class wireless LANs. Virtualizing the controller has some real benefits in distributed-data implementations.
I recently spent some time at Bluesocket learning about their recently-announced vWLAN architecture. Bluesocket has been a major player in enterprise-class WLANs essentially since the beginning of the “switched” WLAN era, which, of course, quickly gave rise to the concept of a WLAN controller. The idea behind the controller is simple – move global system control functions to a central location. This concept remains controversial, with some vendors claiming that a distributed control model with the control plane resident in each AP is the right way to go, and, as we discussed last time, the concept of a moveable, relocateable control plane also taking root. And of course, there’s the issue of whether user data should flow through the controller. We’re not going to settle these issues anytime soon, again for the reasons I discussed in my last entry here.
But suppose you have a controller-based architecture (as most enterprise-class installations do), and suppose your architecture also includes direct forwarding, in which user traffic doesn’t need to flow through that controller. Wouldn’t it make sense to virtualize the controller functionally, reducing it to software that can run in a virtual machine? Duty cycles and traffic flows are both light enough so that concerns about performance are minimal (and controllers, after all, are essentially single-board computers, often running Linux, so running this code on a powerful server using VMware or a similar hypervisor should work just fine). It’s easier with this approach to implement fault tolerance, likely significantly less expensive than using dedicated hardware, and going the virtual route has inherent flexibility for both vendors and network operators that can’t be matched with hardware-based solutions.
If there’s any downside to this strategy, I’m not seeing it, and I think Bluesocket’s approach will become broadly influential over time. You can read the details in our latest White Paper on this subject. I’ve been promised access to a customer for some performance analysis in the near future (these products are just rolling into production now), so I’ll have more shortly.
It’s also important to point out that the term “virtual” in the context of a WLAN has a number of other meanings as well – here we’re just talking about an implementation strategy that changes the cost curve and really reflects the overall trend to server virtualization in the enterprise. I will continue the discussion of virtualization in a broader context shortly.
