Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Community Member

WLC Campus Design - 10 controllers etc

Hi Guys,

If I have the following scenarios,

2 sites and each site will be dispersed by 100km (just an example to show that there is no wireless reach between them) and have 100 APs in each site.

Please see diagram attached.

Both sites either have two controllers on site or in a centralised DC (if there a preferred method)???

Anyway,

Two controllers for site1 (1 at each DC for instance) should be in the same mobility group.

Should the site 2 controllers be in a different mobility group?

Also, if this design expanded by a factor of 5, and we now had 20+ controllers, should the controllers each be on a seperate VLAN in the DCs or could I create a Wireless controller VLAN that has all the controllers connected to?

Please note that on the previous point, the DCs have NO layer 2 VLANs between them, so each set of controllers (per site, one in each DC) are susceptible to Layer 3 roams, but the rule of thumb we will use is that each site will have a primary controller and a secondary controller and all APs should be on the same controller.

Also, for smaller Wireless deployments, is it best to have a set of controllers that nmanager small sites, say for 5-10 APs per site? BUT then, you can only have one mobility group for the controller so would this cause a problems? Oh, now im confused?

Can anyone comment on the designs above?

Many thx,

Ken

1 ACCEPTED SOLUTION

Accepted Solutions
Hall of Fame Super Silver

Re: WLC Campus Design - 10 controllers etc

H-REAP works well in the 4.2 code. I have about 30 sites with up to 5 1252's running in H-REAP mode back to the a single DC. The DC has 2 4404-100's and I split up the sites to be primary on one and secondary on the other.

As for redundancy, you have to look at how will you manage the ip addressing of clients. N+N is the most popular because you have still 100% redundancy. This either can be at the site or back at the DC.

Now you can also do N+1, one in the local site and the other in the DC. Now you have to look at what ip address users will get when they are on the primary wlc and when the primary fails and the ap's go to the backup wlc. Now you have clients getting and ip address on the central site (DC) and terminating traffic there. So now you will have to look at traffic flow and how will you configure your acl's if required.

N+N+1 Can work for you if you have a WiSM's in both DC. Then for site 1:

Primary DC1WISM1-WLC1

Secondary DC2WISM1-WLC1

Tertiary DC1WISM1-WLC2

Or you can have one 4404-100 at each site as primary and a 4404-100 at each DC as secondary and tertiary. You will need one at each site and two at each DC just for your two big remote sites. Now you have to figure out what is required for the smaller remote sites.

It comes down to cost and is it best to keep the wlc's on site since if the wan goes down, resources can't be reached at the DC. Also If you place the primary and secondary and or tertiary in different ip subnets, then how will you manage ip addressing client will get from the various wlc's.

Also.... Stick with the 4.2 if you need to deploy h-reap. Don't go with the 5.0 or the new 5.1. %.0 is the worse and 5.1 is too new and might carry over the issues in the 5.0 and might introduce more.

-Scott
*** Please rate helpful posts ***
3 REPLIES
Hall of Fame Super Silver

Re: WLC Campus Design - 10 controllers etc

I will give this a shot....

First, if most of the resources are in the DC's then having the WLC's at the DC's is the best design. The reason I say this, is because most organizations are trying to centralize resources than having resources at every site. Now.. what is the link from the site to the DC's? If you have enough bandwidth, then you can run 100 ap's or so in local mode. If you don't have enough bandwidth, then look into designing the wlan with h-reap ap's. For the mobility group, just keep it different. For the smaller site, it doesn't matter if you have the same mobility group.... because if you wanted to, then you would be buying 2106 or 2125's or 4402-25 for each site. I have a huge h-reap deployment with 30 sites and all h-reap ap's come back to two wlcs with the same mobility group. As for subnet, I would make the subnet small /29 and keep the site 1 and site 2 on different subnets.... This way you can avoid any ap's finding the wrong wlc.

Hope this doesn't confuse you.

-Scott
*** Please rate helpful posts ***
Community Member

Re: WLC Campus Design - 10 controllers etc

Fella 5 (another 5 coming up)

That is most helpful. The major sites all have enough BW to the centralised DCs - as you state, most companies are looking at centralised DCs with users site being "thin", ie no servers or a little servers as possible.

Also you mention h-reap. I need to read up on this. This could be good for sites that have APs and have to travese a WAN to get to a WLC. Have you ever done this? Gonna read up now.

As for the subnets, currently we have four WLCs (2 in each DC) and 1 in each DC is for a certain site (100APs +) but unfortuntley, they are all on the same subnet in each DC, different subnets at each DC, ie, two on subnet A in DC1 and two on subnet B in DC2. Is this a real show stopper and should I move them for sure onto seperate VLANs at the DC because this will cause issues.

One last point, is that we donot have L2 VLANs between the DCs. It is all L3 routed, so we try to keep all APs for a site on the same WLC and use the second WLC as a backup to minimise L3 roams. We use a N+N redundancy deployment. What are your views on N+N, N+1 and N+N+1? and have you heard of any other WLC reduncancy designs?

You help is very valued on this subject.

Many thx and kind regards,

Ken

Hall of Fame Super Silver

Re: WLC Campus Design - 10 controllers etc

H-REAP works well in the 4.2 code. I have about 30 sites with up to 5 1252's running in H-REAP mode back to the a single DC. The DC has 2 4404-100's and I split up the sites to be primary on one and secondary on the other.

As for redundancy, you have to look at how will you manage the ip addressing of clients. N+N is the most popular because you have still 100% redundancy. This either can be at the site or back at the DC.

Now you can also do N+1, one in the local site and the other in the DC. Now you have to look at what ip address users will get when they are on the primary wlc and when the primary fails and the ap's go to the backup wlc. Now you have clients getting and ip address on the central site (DC) and terminating traffic there. So now you will have to look at traffic flow and how will you configure your acl's if required.

N+N+1 Can work for you if you have a WiSM's in both DC. Then for site 1:

Primary DC1WISM1-WLC1

Secondary DC2WISM1-WLC1

Tertiary DC1WISM1-WLC2

Or you can have one 4404-100 at each site as primary and a 4404-100 at each DC as secondary and tertiary. You will need one at each site and two at each DC just for your two big remote sites. Now you have to figure out what is required for the smaller remote sites.

It comes down to cost and is it best to keep the wlc's on site since if the wan goes down, resources can't be reached at the DC. Also If you place the primary and secondary and or tertiary in different ip subnets, then how will you manage ip addressing client will get from the various wlc's.

Also.... Stick with the 4.2 if you need to deploy h-reap. Don't go with the 5.0 or the new 5.1. %.0 is the worse and 5.1 is too new and might carry over the issues in the 5.0 and might introduce more.

-Scott
*** Please rate helpful posts ***
177
Views
5
Helpful
3
Replies
CreatePlease to create content