Roaming between Flexconnect groups for scaling

[kbyrd]

I have a customer that needs flexconnect at each of his 10 locations to access local servers and printers. The customer has a pair of 5508 WLCs running 7.6.130.0.

While the customer currently has 25 and under AP count per site, they are considering an expansion to 50 - 60 per site.

We are considering the mobility agent on 3650/3850/4500 switches, but the multi-hop restriction will drive the cost too high.

What is the downside for defining multiple flexconnect groups per site?

The customer is also considering Unified Communications. For example, would the voice RTP stream on a wireless IP phone roaming between APs on different flexconnect groups appear to be seamless?

[David Watkins]

If you plan on utilizing any real-time applications such as voice, you would not want these devices to be roaming between FlexConnect Groups.  There will be a full re-authentication of the client; with the exception of OKC capable machines, which "may" roam more cleanly.  This means some standard data clients may perform a fast roam, or at least not notice much of a hiccup even with a full re-auth. 

In either scenario, you would want to make sure this is NOT a L3 mobility roam (ie. FlexConnect WLAN/VLAN mapping to different networks).  This will cause major problems for all your clients as they will most likely end up talking on the new VLAN with their old IP address.

Mobility / Roaming Scenarios

WLAN Configuration	Local Switching			Central Switching
	CCKM	PMK (OKC)	Others	CCKM	PMK (OKC)	Others
Mobility Between Same Flex Group	Fast Roam(1)	Fast Roam(1)	Full Auth(1)	Fast Roam	Fast Roam	Full Auth
Mobility Between Different Flex Group	Full Auth(1)	Fast Roam(1)	Full Auth(1)	Full Auth	Fast Roam	Full Auth
Inter Controller Mobility	N/A	N/A	N/A	Full Auth	Fast Roam	Full Auth
(1) Provided WLAN is mapped to the same VLAN (same subnet).

[iores]

@David Watkins Could you please provide link with the table above?