Wireless Access

I'm looking for some design advise on how to add additional controllers.  Our topology is two 3600 controllers serving as Master/Master-backup using VRRP (No APs will terminate on the master).  Two 6000 chassis with one M3 in each.  The two M3 controllers are layer 2 connected at 1Gig and configured with two VRRP instances between them.  The 1Gig connection on the two M3 controllers is being upgraded to 10Gig and four additional M3 controllers will be added at 10Gig as well.  This will bring the number of M3 controllers in each 6000 chassis to three.  Our goal is complete redundancy and the two 6000 chassis are located in different hubs on the network.  We will have approximately 2000 AP134's operating between the six M3 controllers.  All M3 controllers will be half loaded, but licensed for max number of APs.

The scaling question is:

1. Should the M3 controllers be added in redundant pairs (one in each 6000 chassis) on separate vlans with their own VRRP instances for redundancy. 
2. Should the M3 controllers be added in redundant pairs (one in each 6000 chassis) in the same vlan with their own VRRP instances for it's pair.

Would the 10Gig connection be sufficient for the traffic generated if all VRRP instances were in the same vlan?  It seems that having all the controllers in the same vlan would enhance client roaming between APs that may reside on different controllers.  All APs in any given building will point to the same local controller (VRRP instance).  I was just concerned about adjacent buildings that may happen to reside on different controllers and roaming between the two.

Thanks in advance

Please look at the following design guideline and let us know if this fits your requirement for redundancy 

6000 Chassis6000 Chassis

Master Standby
Local1 Local2
Local3Local4

1. No APs would terminate on the Master / Standby pair
2. 500 APs on each local controller
3. 2 VRRP instances between Local1 and Local2 with VRRP IPs as VIP1 and VIP2
4. 2 VRRP instances between Local3 and Local4 with VRRP IPs as VIP3 and VIP4
5. Local1 will own VIP1 and act as STANDBY for VIP2
6. Local2 will own VIP2 and act as STANDBY for VIP1
7. Local3 will own VIP3 and act as STANDBY for VIP4
8. Local4 will own VIP4 and act as STANDBY for VIP3
9. 4 AP-Groups with 4 different AP System Profiles
10.AP SysProfile1 - LMS - VIP1, Backup LMS - VIP2
11.AP SysProfile2 - LMS - VIP2, Backup LMS - VIP1
12.AP SysProfile3 - LMS - VIP3, Backup LMS - VIP4
13.AP SysProfile4 - LMS - VIP4, Backup LMS - VIP3
14. Assign 500 APs in each AP-GROUP

The Campus VRD available here also provides a good design guideline that will apply to your network requirement. 

Regards,

--
HT 

Thanks for the feedback, but it did not answer the specific question regarding all the VRRP instances living in the same vlan.  We have already implemented the solution of having a separate vlan for each redundant pair of M3 controllers.

It is OK to have the VRRP instances in the same VLAN.