DHCP loadsharing with redundant Guest Anchor Controllers

[eshrarosman]

Hi

I have 2 x Redundant Guest Anchor Controllers (5508) located in 2 separate Data Centres with all the management and guest user VLAN spanned between two. Everything is working fine with the Guest WiFi access except the DHCP functionality as the Controllers are acting themselves as the internal DHCP Servers.

This is how I tried to distribute

network. 10.1.0.0/23

gateway: 10.1.1.254

Controller 1, DHCP Server pool: 10.1.0.2 - 10.1.0.254 Gw: 10.1.1.254

Controller 2, DHCP Server pool: 10.1.1.2 - 10.1.1.254 Gw: 10.1.1.254

As the user loadbalancing between the Anchor Controllers cannot be controlled (i.e. they are active/active), the same client sometime getting 2 different IP addresses from both the Controllers (as they do not talk to each other in terms of DHCP) hence depleting the pool addresses.

I guess one way of solving this is to just run 1 DHCP server in one of the controllers but that defeats the purpose of having N+1 Controllers. Is there a better way of doing the DHCP loadbalancing and having full redundancy at the same time?

Any suggestion will be greatly appreciated.

Regards

[Scott Fella]

Sometimes it's just better to put a dhcp server in the dmz and have both wlc point to that. If you have mobility setup right, if a client roams from one wlc ap to another wlc ap, they should not have to dhcp again. I don't know why you are seeing a guest client with two address, but the wlc isn't that great of a dhcp server also.

Sent from Cisco Technical Support iPhone App

[eshrarosman]

Thanks Scott, I understand that it's quite obvious to get an external DHCP Server, unfortunately it's not an option for us The weired thing is, it seems when a client joins the guest WiFi, both the Anchor Controllers (both functioning as DHCP servers with mutually exclusive IP Address space) are providing IP addresses. While the client accepts only one the other Controller still reserves the IP address unused and hence depleting the DHCP Pool.

I thought for load balancing (in the very beginning) the Foreign controller will forward the DHCP request to only one of tthe Anchor Controllers, but in reality it's forwarding it to both. I have tested this with only one test AP, so mobility doesn't seem to be an issue here. Any thoughts?

[Michael Odom]

We have the same configuration in our organization, the best solution I have found is to set the DHCP lease time to something small, like 15 minutes or less.  It's not exactly an elegant solution, but it also helps free up addresses  from smartphones that briefly connect and then leave.