Is it considered best practice to use the internal DHCP servers on the controllers?
I am experiencing a rather strange issue where I get seemingly random disassociations on my WiSMs.
I am running a WiSM on software 188.8.131.52 and WCS version 184.108.40.206. My LWAPP network is consisting of 1010 and a few 1020 APs.
I may have traced the problem down to the fact that some access points have duplicate IP addresses.
For the Access Points, the DHCP server is my Core 6509, running IOS 12.2(18)SXF12. I feel this may be the source of the issue and would appreciate any input on how this can be prevented- it seems like the 6509 does -not- check that an IP exists before handing them out.
I personally would choose option #3. Use the DHCP solution you use for clients. Typically networks have a robust solution in place for DHCP for their clients, and the same solution is often suitable for APs as well.
I avoid the IOS DHCP server unless it is the only option. I prefer to let the router route, and let a more functional DHCP server serve up IPs.
The same goes for the controller DHCP server. Using the controller has some issues, particularly if you have multiple controllers.
fella: I'm curious your reasons for static IPs on the APs. Care to elaborate? I have done many deployments, and everywhere that has requested to do static has quickly changed their minds when they see the challenges of working with the APs when they move them. Moving a static AP from one building to another requires a lot of work vs. just plugging it in on a different DHCP segment. For very small deployments it can be workable, but large deployments I don't find it is.
IMO, one of the biggest benefits of the controller methodology is the AP IP is largely irrelevant, as long as it can get to the controller. I do suggest separating the APs IPs from user IPs just to avoid IP conflicts if users hardcode their IP.
I would agree, set the AP's to use DHCP and forget about them.
Use your DHCP server (Windows, Network Registrar, etc..) to hand out DHCP addresses to your clients. Now for the Wism modules, I would use DHCP for their management at first to configure them, and then set them to a static IP address (then blow away the DHCP pool on the 6500). Something about using DHCP for management makes me un-easy.
I have always used static ip address on the ap. For one, there should be no reason to move an ap to another subnet unless you have a whole bunch of ap's that died. If you need to move one, all you have to do is change the static ip to one in that subnet and hit apply. Then remove the ap and plug it into the other subnet. Easy than going in the dhcp and adding a reservation and removing th eold one. Also look at the cycle when an ap joins a controller it has to broadcast for a dhcp. With a static IP, it converges a bit faster than if it has to obtain an ip.
It's easier to manage static ips than creating reservations.... I Think.