Forum

http://www.networkworld.com/news/2006/100906-trapeze.html

Trapeze seems to be adopting the hybrid architecture whereby the controller is removed from the data plane. This is GOOD.

Colubris adopted this architecture in 4.1, but in the 5.1 beta I'm testing, it seems they have a switchable platform whereby you can have the "centralized" architecture or autonomous APs in a hybrid architecture. Odd.

It's interesting to see which companies take which stances in this technology mashup called Wi-Fi.

Devinator

It is an interesting thought. 802.11n and a centralized controller is going to be tough to do, so using an architecture like this is a good idea. So, standard bulk data goes from the AP to the LAN, and the VoIP traffic goes to the controller to reduce latency and improve roaming. Does this sound correct?

Had the chance to attend a mobile roadshow with one of the Trapeze reps there. I was impressed by the design methodology and less complicated methods of setting ip a network using the GUI. Ring Master is SWEET!
Their design will make the job of the "Mobile Access Specialists" easier and exciting.

With their Fixed Mobile Convergence (FMC) strategies all that is left is a combination of cognitive and multi-band access radios in the access portals that progate 3G to 4G signals (Cellular-WI-FI-WiMAX) like the Xirrus device. Small form factor -ethernet based with meshing capabilities?

Hi Gene,

You had the first part right, but with wVoIP, the latency of encapsulation/decapsulation going to/from the controller is a bad thing...especially if the bulk data is ALSO going with it and QoS isn't as good as it should be. It's best to implement QoS at the AP and have

Having said that, Colubris's QoS is remarkable to say the very least - the best I've seen in fact, but the controller is only part of the picture when you're tunneling everything to the controller. You have to take into consideration the upstream switches and routers between the AP and controller also.

Devin

Cisco do something similar, they call it H-REAP. Cisco's slant on it is that you use it for remote sites with limited WAN bandwidth. Using H-REAP, the controller (middle of your WAN) determines all of the ESSID names, characteristics, etc..., and what VLANs the SSIDs map to, where that VLAN is, and how authentication works, etc...

Net result is that you can have SSIDs that can be set to use;

Remote Site Breakout
or
Central Site Breakout
or
Central Site Breakout, failing back to Remote Site Breakout if WAN fails.

And obviously you can specify QoS (802.11e & 802.1p) for each SSID as required. Also worth noting that when running VoIP, there are only four levels of "wireless-QoS" compared to six levels of "wired-QoS", so some re-mapping does occur...

Rich