Forum

Greetings fellow CWNPers having been out of wireless for awhile I have this new project and I have a question for the group and was curious if anyone has done this. I am looking through the docs on CCO and have gone through several books:

Controller-Based Wireless LAN Fundamentals ? very good and updated 2010 publication and Deploying and Troubleshooting Cisco Wireless LAN Controllers also very good but from 2009.

I have a new install coming up over a legacy install. What I mean is we have a classic autonomous AP based setup on the campus with one vlan. We now want to deploy 5508 controllers and new N based AP. My questions are for migration approach. One school of thought is to deploy the controllers and convert all the autonomous fat APs to LWAPs then replace each LWAPs with a new N unit one or x at a time at our convenience. Then convert the removed FAT to converted LWAPs back to FAT (IOS) for another special classic non controller use.

I was curious if I can save this conversion step and deploy the controllers thus letting the users continue to use their classical FAT architecture while I add the N based APs one at a time. So, for example, I have the fat architecture now, deploy controllers, and on one floor where there is 5 fat APs swap one with an N AP. Now anyone in that N cell range will go through the controller but the other folks in the other cells can continue to authenticate and associate using the fat system in place. Can the controllers and Autonomous systems reside together? The only immediate protocol mechanics related issue here is roaming. I want to avoid any issues related to converting the FAT aps to LW aps in production(bugs, rollbacks etc that can slow things down). This way I can get the controllers and new APs in place quickly utilizing some known RF variables and slowly push the older APs out of use. Kinda like a SIN routing protocol migration flip with Admin distance flipping.

Is this possible and has anyone done this? Just curious. I am still looking though the docs. Any thoughts are greatly appreciated.

I cannot wait to get the new CWAP and CWDP books.

I'm a bit confuse with the description of your scenario. please correct me if i'm wrong. The way I understand it, you want to upgrade your legcy 802.11 network to a controller based 802.11n network. Its that right?

If yes, I would suggest the following:
1. Determine first your client stations and type of applications running. How many are using "bg", "a", "n"? It is very important to know them prior to deciding which type of equipment you are going to purchase. This will save you a great deal of money.

2. Do a site survey. If you have a predictive planning software, it is good as it will give you an indication of the AP placement, number of APs, type of APs, etc... But if you want to get a more accurate site survey, do a "active site survey". This manual site survey will give you a better results of the site survey parameters that you need but is very time consuming compared to predictive planning.

3. Put in mind the type of power injectors if you are going to use it. This is important because 802.11n APs consumes more power than the abg APs.

4. Base on my experience, it is a good practice to have a somewhat perfect deployment and to have your required config(VLANS, portals, security) on the 2nd phase of your project.

I wish you all the best on your upcoming project and I hope this reply can help in some ways.

From a coverage perspective, what was good for "b/g" may not be good for "n". So a one-for-one replacement may not be the best choice.

MIMO "uses" multipath. The best a b/g system with switched diversity antennas can do is only try to "compensate" for it.

If you are eventually going to phase out the older infrastructure, I would do an "n" site survey first.

Thanks guys, and yes that is basically it. I want to phase out the older APs and try to save a step(convert old APs to lwap and then back). I do understand about the cell coverage issues and that just swapping N for G will introduce holes/gaps in area. However, according to the books some of the N ap will keep the same coverage area. The new equipment has been ordered already. I don't know the models yet for old and new other than 5508 controllers so I understand all the issues with site surveys and possibly duel band usage plus the client radios as well. I don't know the client population and uses on the existing system yet until I am on the ground. I am trying to nail down an approach to get the new units out if the existing uses are minimal and roaming is not a big deal. This way I can get the controllers, in and replace an N ofr older one then when all the Ns are in adjust for coverage etc. To start with the survey is tough(I may have to do it anyway) for the client is fixed on using the spots where the older units are. Thus the survey may change all of that. A tricky one it is.
Thanks for the ideas and any other thought would be great.

Having done a similar upgrade, yes the autonomous b/g APs and LWAPP n APs can exist together and roaming works between the two, although not as fast as if it were all LWAPP (depending on security in place). I would recommend manually setting the power settings on the LWAPP APs until they're all replaced at least. We found that the 11n APs had better range for 11g clients due to beamforming and MRC.