Wireless Access

Hi all,

 

I have a question about clustering in AOS8, version 8.2.1.0.

 

If I create a cluster of 2x 7210 MC's, will I have AP capacity of 1024 AP's or 512 AP's?

This is assuming I want full AP and client redundancy.

 

Will each controller support 512 active tunnels and 512 standby tunnels of just 256 active tunnels and 256 standby tunnels?

 

Adding a 7220 controller to this cluster, will this then increase my AP capacity by 1024 or 512 AP's?

 

Regards

 

Hey Micky,

 

For Homogeneous cluster, Use the formula "Cluster AP count = 50% cluster capacity"
7210 controller have a capacity of 512 APs, meaning that a cluster of 2 X 7210 has a combined capacity of 1024 APs.
In order to ensure that every AP has an AAC and S-AAC with adequate capacity for all APs to failover the recommended AP load of this cluster should be half of the total cluster capacity.
So Cluster AP size is 512 APs.

 

For Heterogeneous Cluster, Use the formula "Cluster AP size = Lowest value of (50% cluster capacity, Worst case scenario load) "
In you example: 2 X 7210 and 1 X 7220
50% of total capacity = (512 + 512 + 1024)/2 = 1024 APs.
Worst case scenario load = (512 + 512) = 1024 APs. [Note: Worst case scenario load is the AP load handled by the remaining nodes in the cluster in the event of highest capacity cluster member going down]
So Cluster AP size is 1024 APs

 

Example to demonstrate worst case scenario load:
A cluster comprised of 1 X 7210, 1 X 7220, and 1 X 7240
50% of total cluster capacity = (512 + 1024 + 2048)/2 = 3584/2 = 1792 APs
The worst case scenario load = (512 + 1024) = 1536 APs
So Cluster AP size = 1536 APs

 

Regards,

Kapil

 

Kapil,

 

Thanks for your reply.

 

This helps me out a lot in planning my design.

 

Regards,

Micky

Welcome Micky

Hi Kapil,

 

if I just want load balance the APs and risk to loose some APs in case of a failure of one controller, can I connect more APs as the platform supports?

For example, 2x 7030 (up to 64 APs) clustered and about 80 APs connected to the cluster.

 

Thanks and Regards,

Uli

Hello Uli,

 

7030 supports 64 APs.
For 2 X 7030 = (64 + 64)/2 = 64 APs is the Cluster AP size.
If 80 APs were used, not every AP will have an S-AAC in normal operation since 80 is greater than half of the cluster's AP capacity.
For normal cluster operations, every AP terminated on the cluster should have an AAC and S-AAC.

 

Regards,

Kapil

Hello Kapil,

 

does that mean, that it is not working correctly at all or that some of the cluster functions like live update won't work (what is obviously ;-))

 

Regards,

Uli

Just a Quick question to be sure....
2 X 7030 = (64 + 64)/2 = 64 APs is the Cluster AP size.

Is it possible to terminale like 80 aps on the cluster and does the controller break the backup tunnel by itself? We have a situation when we add a third controller later. We want to get all aps online.

Thanks

You will have some number of APs that will not have a standby tunnel. Later on when you add a 3rd controller, the cluster manager will re-balance. 

Does that mean we can connect 80 AP's on 2X7030 controllers but in the event if one controller goes down, only 64 will come up on remaining controller?

That is correct. If a 3rd cluster member is added down the road, the cluster leader will reallocate the APs and the users across the cluster members. The key is, if an MC fails, can the other MCs support all of the APs. If the answer is yes, then you are good. If not, then if an MC fails, some of the APs will not be able to establish tunnels with the remaining MCs and will not come up. You do not have control over which APs do or don't come up.

I hope this helps,

 

Thank you for your reply that brings up another question.

Lets say we are running 2X7030 controllers in a cluster with 80 AP's and 80 required licenses. Assuming that AP load balancing will distribute the AP's on those two controllers. Let's assume we ended up with 50 AP's on controller1 as their active controller and controller2 as their standby controller. For remaining 30 AP's controller2 as their Active controller and controller1 as their standby controller. which means all 80 AP's has active and standby tunnels already established.

 

Now if controller1 goes down will those 50 AP's move to standby tunnel and continue working on standby controller(controller2 in our case), until they reboot or only 34 AP's will move to the standby controller and then standby controller won't accept other AP's?

I'm not sure of the exact mechanics, but the bottom line would be that the 30 APs that were/are connected to MC2 would stay there. Of the 50 APs that were connected to MC1, only 34 would be accepted on MC2.

I hope this helps,

 

---

@ashwani.kaundal wrote:

Thank you for your reply that brings up another question.

Lets say we are running 2X7030 controllers in a cluster with 80 AP's and 80 required licenses. Assuming that AP load balancing will distribute the AP's on those two controllers. Let's assume we ended up with 50 AP's on controller1 as their active controller and controller2 as their standby controller. For remaining 30 AP's controller2 as their Active controller and controller1 as their standby controller. which means all 80 AP's has active and standby tunnels already established.

 

Now if controller1 goes down will those 50 AP's move to standby tunnel and continue working on standby controller(controller2 in our case), until they reboot or only 34 AP's will move to the standby controller and then standby controller won't accept other AP's?

---

You won't have 80 tunnels (active and standby) to a single controller (assuming this model controller). In your example with both controllers active, some of the APs will not have a standby tunnel established because the controller capacity is exceeded. If their primary controller failed, the APs without a standby tunnel would go offline.