Hi
We are facing a design issue when having almost 10.000 Ap's and still growing.
At det moment we have set up Cluster 1 and running 3500 aps, and for our understanding talking with several aruba engineers its not possible for airwave to handle more then 4048 Ap's. This leaves us with 1 Cluster = 1 airwave.
So we would end up with 3 clusters for controlling up to 12.000 ap's
Cluster = 4 x 7240xm (2 in DC1 and 2 in DC2). One DC or 2 controllers should be able to handle all 4048 ap's in case 2 controllers goes down.
--Managed Networks
-----MCG
--------Cluster 1
--------Cluster 2
--------Cluster 3
The Main problem, and where the design issues lays, is how do we administrate how the AP's finds its cluster the best way.
As we see it, we have 3 choices:
1. We control the master ip/vrrp of the cluster on the DHCP scoopes in option 43. (ber in mind we have 900 locations) we would have to do a static plan, where we divide the locations in 3. this would't be a very dynamic solution.
pros.
- No ekstra configurations on the mobility master, and we are still able to inherit configurations from a top level and use i across all clusters.
- No cluster can't be overrunned with ap's per design, because of the division.
cons.
- The administrative headache in maintaining a list where the ap is connecting and the fact that we are letting the dhcp control where our ap should go.
- static configuration on each scopes and the need of a maintained plan for documentation to make sure the cluster is not taking in to many ap's.
2. We use option 43 from DHCP and configure cluster 1's vrrp, like we are doing today. When the AP is hitting cluster 1 we can control where it should go through LMS from the AP-group. This means we need to create and configure a specific ap group with an LMS ip, one for each of the 3 clusters under the ap system profile in the MCG folder.
pros.
- we control where the ap connect from the mobility master and in the controller environment.
cons.
- We have 25 ap groups and more to come, so this would mean we need 3 x 25 = 75 ap groups for all 3 clusters to handel all the solutions we have. And every time we make a change we should maintain 3 ap groups.
- Potentially cluster 1 could be overrunned if cluster 2 and 3 is unavailable and goes down, because of the the default behavior in discovering the master through DHCP. But then you probably have bigger problems on hand.
3. We use option 43 from DHCP and configure cluster 1's vrrp, like we are doing today. we then configure each ap with a static master.
pros.
- we control where the ap connect from the mobility master and in the controller environment.
- No ekstra configurations on the mobility master, and we are still able to inherit configurations from a top level and use the same ap groups across all clusters.
cons.
- Potentially cluster 1 could be overrunned if cluster 2 and 3 is unavailable and goes down.
No offens, but we are really hoping that someone in this community is in the same situation as we are. We are really struggling to find the right design, that would scale to more than 10.000 aps taking into account the limitation that airwave have.