Wireless Access

Hello, I have a question regarding Aruba OS 8.

 

If I cluster together two controllers running AOS8, each supporting 512 APs and one of them fails, will the other controller take over the extra 512 APs with over subscription for the time it takes to change the faulty controller?

 

In ArubaOS 6.5 this function existed, and I was just wondering if this function is still available in AOS8. It is an inportant factor when I upgrading.

The oversubscription in AOS 6.x is for the standby tunnels when using HA

In AOS 8 Since each AP in cluster will have a AAC and S-AAC , I assume oversubscription for standby tunnels also is taken in consideration

Hi, would a virtual controller also support oversubscription?

In what context?

Hi Colin

 

Well, with OS6 HA a 72xx controller can terminate 4 x it's capacity of backup tunnels. Can a VM50 be a Standby AAC for more than 50 AP's?

With clustering "oversubscription" causes standby tunnels to be dropped in favor of active tunnels regardless if you are using VMCs or Hardware Controllers.  If an access point comes up on a controller that is at "standby" capacity, one standby tunnel is dropped to allow that AP to come up.  For example, if an entire controller fails and that puts you systemwide at more than 50% AP capacity, standby tunnels will be removed to allow active AP tunnels to establish.

 

I hope that makes sense.

Yes that makes sense. I understand that the standby tunnels will drop when needed. However, I think what I'm looking for is the 'standby capacity' for the VMC's (?) A 72xx can serve up to 4 x it's active capacity for standby tunnels if I'm not mistaking. With a 70xx there is no 'oversubscription' possible. But what about VMC? Let's say we have 2 x 7010's with 64 AP's. At that point they are at their capacity. What VMC should I add to have Clustering HA? VA50 or VA250?

The "4x its standy capacity" is really only for HA in 6.x, where redundancy is configured manually.  The 4x number is for planning.  In 8.x, there is no "4x capacity" concept, because redundancy is done dynamically.

 

You should only have the same model of controllers in a cluster to avoid any issues with attempting to compute what access points will have redundancy. 

 

If you only plan on 50% AP capacity in a cluster, you will have full redundancy.  If you are at more than 50% APs capacity in a cluster, needless to say, some APs will not have full redundancy if a controller fails.

 

If you haven't, please see the cluster manager webinar here:  https://community.arubanetworks.com/t5/Wireless-Access/Technical-Webinar-Recording-Slides-ArubaOS-Cluster-Manager/m-p/297761?author_id=35349&collapse_discussion=true&filter=authorId&page=2&q=technical%20webinar&search_type=thread

 

 

Thx Colin!