WEBVTT

0:00:02.860000 --> 0:00:07.760000
 Hello and welcome to this video titled
 Wireless Land Controller Deployment

0:00:07.760000 --> 0:00:11.960000
 Options. In this video I'm going to be
 talking about Wireless Land Controller

0:00:11.960000 --> 0:00:15.180000
 Deployments. We're going to focus primarily
 on things called the centralized

0:00:15.180000 --> 0:00:17.840000
 wireless network architecture.

0:00:17.840000 --> 0:00:21.280000
 I'm going to talk about something called
 FlexConnect, which is a way to

0:00:21.280000 --> 0:00:24.580000
 make the centralized wireless network
 architecture work a little bit better

0:00:24.580000 --> 0:00:26.680000
 in some situations.

0:00:26.680000 --> 0:00:30.560000
 We're also going to talk about the converged
 wireless network architecture.

0:00:30.560000 --> 0:00:33.240000
 By the time we're done here, you'll
 understand the difference between

0:00:33.240000 --> 0:00:36.420000
 all three of these.

0:00:36.420000 --> 0:00:41.700000
 In this video we're primarily focusing
 on on-site deployment of wireless

0:00:41.700000 --> 0:00:43.140000
 land controllers.

0:00:43.140000 --> 0:00:46.560000
 As you can see here, there's two
 methods or two options for this.

0:00:46.560000 --> 0:00:50.620000
 You can do what's called centralized, or
 you can do a converged architecture.

0:00:50.620000 --> 0:00:55.620000
 Like the last bullet point says, each
 architecture has a different approach

0:00:55.620000 --> 0:00:59.680000
 to how data frames
 will be forwarded.

0:00:59.680000 --> 0:01:02.740000
 Let's start with a centralized
 architecture.

0:01:02.740000 --> 0:01:07.460000
 This typically means that you have one,
 maybe two controllers placed in

0:01:07.460000 --> 0:01:09.440000
 a central location.

0:01:09.440000 --> 0:01:12.940000
 That central location, for example, could
 be in your data center somewhere,

0:01:12.940000 --> 0:01:16.680000
 but a central location of your company,
 and all your wireless access points,

0:01:16.680000 --> 0:01:20.040000
 whether they're within that building,
 other buildings, or even at remote

0:01:20.040000 --> 0:01:24.380000
 sites connected via a WAN or internet
 connection, have to come back and

0:01:24.380000 --> 0:01:30.080000
 form their CAPWAP tunnels with that
 one or two centralized controllers.

0:01:30.080000 --> 0:01:35.500000
 Now this maximizes the quantity of
 APs each controller can control.

0:01:35.500000 --> 0:01:39.100000
 So clearly to do this, we're going to
 need controllers that can support

0:01:39.100000 --> 0:01:44.400000
 more quantities of access points, that
 have more memory, more CPU power,

0:01:44.400000 --> 0:01:46.320000
 so they can handle this.

0:01:46.320000 --> 0:01:51.700000
 Now the positive side of this is that if
 all of my access points are connected,

0:01:51.700000 --> 0:01:55.940000
 or I should say managed by one controller
 in a data center, I've got full

0:01:55.940000 --> 0:01:58.100000
 visibility to everything.

0:01:58.100000 --> 0:02:02.240000
 Once I log into the GUI of that particular
 controller, I have full visibility

0:02:02.240000 --> 0:02:04.700000
 to my entire Wi-Fi environment.

0:02:04.700000 --> 0:02:08.120000
 I don't have to jump from one controller
 to another, or Heaven's forbid,

0:02:08.120000 --> 0:02:12.560000
 from one autonomous access point to
 another to actually see what's going

0:02:12.560000 --> 0:02:14.640000
 on in my Wi-Fi environment.

0:02:14.640000 --> 0:02:18.080000
 Now here's the thing you've
 got to be familiar with.

0:02:18.080000 --> 0:02:20.820000
 Remember that in this environment we're
 talking about your lightweight

0:02:20.820000 --> 0:02:25.980000
 access points building CAPWAP tunnels,
 two of them, to the controller.

0:02:25.980000 --> 0:02:30.100000
 You've got a CAPWAP tunnel for the management
 and control data, and another

0:02:30.100000 --> 0:02:34.740000
 CAPWAP tunnel for the actual user data
 from two and from the Wi-Fi clients

0:02:34.740000 --> 0:02:39.640000
 themselves. Now, depending on how far
 away that centralized controller

0:02:39.640000 --> 0:02:43.340000
 is from the lightweight access point,
 we might have to be concerned with

0:02:43.340000 --> 0:02:47.200000
 the round-trip delay between those two,
 because traffic is going to have

0:02:47.200000 --> 0:02:49.420000
 to go up that tunnel
 and then back again.

0:02:49.420000 --> 0:02:52.600000
 So as you can see here, the design guidance
 is that the round-trip time

0:02:52.600000 --> 0:02:57.500000
 should be 100 milliseconds or less between
 the lightweight access point

0:02:57.500000 --> 0:03:00.240000
 and that centralized controller.

0:03:00.240000 --> 0:03:03.220000
 So we can see right here, that lightweight
 access point, if we measure

0:03:03.220000 --> 0:03:08.400000
 the time from it to send a frame, any
 kind of frame, data or control to

0:03:08.400000 --> 0:03:11.620000
 that controller, which in this case
 is in the data center, should be 100

0:03:11.620000 --> 0:03:13.780000
 milliseconds or less.

0:03:13.780000 --> 0:03:17.980000
 Now, where you place that controller
 could also be one of two places.

0:03:17.980000 --> 0:03:20.600000
 This is a very popular location right
 here, but you really have to ask

0:03:20.600000 --> 0:03:26.480000
 yourself, okay, of all my Wi-Fi clients
 are connected via Wi-Fi, where

0:03:26.480000 --> 0:03:29.600000
 is most of their data going to go?


0:03:29.600000 --> 0:03:34.080000
 In other words, is most of the data
 from my Wi-Fi clients destined for

0:03:34.080000 --> 0:03:36.520000
 servers in my data center?

0:03:36.520000 --> 0:03:40.500000
 If that's true, then the controller
 should be placed there in the data

0:03:40.500000 --> 0:03:43.460000
 center. Or an alternative
 approach might be this.

0:03:43.460000 --> 0:03:46.620000
 You might say, well, hey, actually, most
 of the destination of my traffic

0:03:46.620000 --> 0:03:50.820000
 is leaving my company and going out to
 the Internet, not necessarily going

0:03:50.820000 --> 0:03:53.020000
 to my own local data
 center right here.

0:03:53.020000 --> 0:03:56.640000
 So in that case, in a centralized wireless
 architecture, you might want

0:03:56.640000 --> 0:04:00.320000
 to have the controller actually place
 closer to your WAN router that has

0:04:00.320000 --> 0:04:02.060000
 your internet connectivity.

0:04:02.060000 --> 0:04:07.280000
 All right, so let's talk about some pros
 and cons of the centralized architecture

0:04:07.280000 --> 0:04:11.320000
 approach. So, benefits.

0:04:11.320000 --> 0:04:15.200000
 By having one single controller placed
 in the data center or close to

0:04:15.200000 --> 0:04:18.580000
 the internet access point, this is
 good because we're going to assume

0:04:18.580000 --> 0:04:23.080000
 that most of your client data would
 need to reach that point anyway.

0:04:23.080000 --> 0:04:25.720000
 So we're just going to tunnel it through
 most of the network because most

0:04:25.720000 --> 0:04:28.820000
 of our network is never going to need
 to see that data, and we're going

0:04:28.820000 --> 0:04:33.200000
 to pop it out of the CAPWAP tunnel at
 the other end real close to where

0:04:33.200000 --> 0:04:36.760000
 that destination is that those clients
 are trying to reach anyway.

0:04:36.760000 --> 0:04:38.620000
 Now, here's the drawback to this.

0:04:38.620000 --> 0:04:41.340000
 And unfortunately,
 there are several.

0:04:41.340000 --> 0:04:45.620000
 Number one, we know that all the wireless
 LAN client data must pass through

0:04:45.620000 --> 0:04:50.020000
 that CAPWAP data tunnel before it can
 be placed natively onto the wired

0:04:50.020000 --> 0:04:54.760000
 network. Now, that might not be a
 good option for several reasons.

0:04:54.760000 --> 0:04:58.880000
 Number one, if that's a long stretch
 of routers and switches we need to

0:04:58.880000 --> 0:05:03.320000
 go through to get from the lightweight
 access point to that controller,

0:05:03.320000 --> 0:05:07.580000
 that could induce a lot of delay of
 that CAPWAP encapsulated traffic.

0:05:07.580000 --> 0:05:10.680000
 And we want to keep that delay
 really, really short.

0:05:10.680000 --> 0:05:13.420000
 Round-trip 100 milliseconds
 or less.

0:05:13.420000 --> 0:05:16.960000
 That's round-trip there
 and back again.

0:05:16.960000 --> 0:05:22.980000
 What if we had, now this doesn't happen
 a lot, but what if we had a situation

0:05:22.980000 --> 0:05:27.940000
 where Wi-Fi clients within the same
 cell, within the same BSS, within

0:05:27.940000 --> 0:05:30.940000
 the same SSID need to
 talk to each other?

0:05:30.940000 --> 0:05:35.220000
 What if John and I need to do some file
 sharing, some peer-to-peer sharing,

0:05:35.220000 --> 0:05:38.360000
 where both wireless connected
 to the same access point?

0:05:38.360000 --> 0:05:39.200000
 Well, guess what?

0:05:39.200000 --> 0:05:43.080000
 Now we have a situation where my traffic
 has to hit the access point,

0:05:43.080000 --> 0:05:47.280000
 go all the way up that CAPWAP data tunnel
 to the controller, where it's

0:05:47.280000 --> 0:05:51.240000
 processed and then shoved back into
 that tunnel all the way down so it

0:05:51.240000 --> 0:05:54.580000
 can pop out the lightweight access
 point and John can see it.

0:05:54.580000 --> 0:05:57.460000
 And now it's got to go all
 the way up and back again.

0:05:57.460000 --> 0:06:04.900000
 So it's not a great option.

0:06:04.900000 --> 0:06:08.300000
 Then they are closer to
 the controller itself.

0:06:08.300000 --> 0:06:13.140000
 So it's definitely not a great option
 for client-to-client communications.

0:06:13.140000 --> 0:06:16.280000
 What about when we have
 branch office laps?

0:06:16.280000 --> 0:06:18.520000
 Branch office lightweight
 access points.

0:06:18.520000 --> 0:06:21.480000
 Well, so we're talking now about a
 situation where an access point in

0:06:21.480000 --> 0:06:35.580000
 a branch office might have to go across
 a slow-speed or even thousands

0:06:35.580000 --> 0:06:38.620000
 of miles away from that lap.

0:06:38.620000 --> 0:06:40.820000
 What happens if that
 tunnel goes down?

0:06:40.820000 --> 0:06:42.420000
 What if we lose our
 internet connection?

0:06:42.420000 --> 0:06:44.240000
 What if we lose our
 WAN connection?

0:06:44.240000 --> 0:06:48.080000
 Well, a lightweight access point
 can't operate on its own.

0:06:48.080000 --> 0:06:49.960000
 It needs a controller.

0:06:49.960000 --> 0:06:53.860000
 So if that tunnel goes down, all those
 people in that branch, they just

0:06:53.860000 --> 0:06:55.320000
 lost their connectivity.

0:06:55.320000 --> 0:06:58.280000
 Because once that lap loses connectivity
 to its controller, it's going

0:06:58.280000 --> 0:07:02.340000
 to disassociate and deauthenticate
 all the clients are connected to it

0:07:02.340000 --> 0:07:06.660000
 because it can't handle
 them all by itself.

0:07:06.660000 --> 0:07:11.220000
 So this could introduce long hairpins
 induced for branch office users

0:07:11.220000 --> 0:07:15.380000
 that need to access resources that are
 actually there in the branch office

0:07:15.380000 --> 0:07:19.840000
 themselves. And like I mentioned, if
 that cap-wap tunnel goes down, no

0:07:19.840000 --> 0:07:23.560000
 Wi-Fi connectivity to anything
 for those branch office users.

0:07:23.560000 --> 0:07:28.100000
 Now for that reason, Cisco developed
 something called FlexConnect.

0:07:28.100000 --> 0:07:30.840000
 Now this used to go
 by previous names.

0:07:30.840000 --> 0:07:34.980000
 It was originally called REAP and then
 hybrid REAP, which stood for remote

0:07:34.980000 --> 0:07:42.260000
-edge access point, remote-edge access
 point, and then hybrid remote-edge

0:07:42.260000 --> 0:07:45.280000
 access point, and then it
 was called FlexConnect.

0:07:45.280000 --> 0:07:48.300000
 But although it's gone through some
 iterations and some upgrades over

0:07:48.300000 --> 0:07:51.940000
 time, the central premise of
 FlexConnect is the same.

0:07:51.940000 --> 0:07:57.960000
 So this feature is a particular mode that's
 available on lightweight access

0:07:57.960000 --> 0:08:01.980000
 points. And the controller
 has to support this.

0:08:01.980000 --> 0:08:06.620000
 It's primarily designed for those remote
-side or branch sites laps that

0:08:06.620000 --> 0:08:09.960000
 are utilizing a controller that's located
 across a WAN connection, or

0:08:09.960000 --> 0:08:13.020000
 across an internet connection located
 at the headquarters, and it's designed

0:08:13.020000 --> 0:08:17.020000
 to alleviate those long hairpins.

0:08:17.020000 --> 0:08:20.820000
 So wireless traffic destined for local
 resources can actually be switched

0:08:20.820000 --> 0:08:23.040000
 locally by the lap.

0:08:23.040000 --> 0:08:28.240000
 In other words, the data does not have
 to go across a data cap-wap tunnel.

0:08:28.240000 --> 0:08:33.660000
 And now this bullet point right here
 really is only referring to if the

0:08:33.660000 --> 0:08:34.840000
 tunnel goes down.

0:08:34.840000 --> 0:08:39.580000
 In other words, if the lap loses its cap
-wap control tunnel to the controller

0:08:39.580000 --> 0:08:41.880000
 and says, Hey, I just
 lost my controller.

0:08:41.880000 --> 0:08:43.000000
 What am I going to do?

0:08:43.000000 --> 0:08:46.860000
 Well, if that access point is operating
 in FlexConnect mode, it can still

0:08:46.860000 --> 0:08:50.480000
 hold on to the access to the clients
 it has, and it could potentially

0:08:50.480000 --> 0:08:54.140000
 even authenticate new clients
 as they come on.

0:08:54.140000 --> 0:09:01.200000
 So when a FlexConnect lap is configured,
 it will operate in one of two

0:09:01.200000 --> 0:09:05.020000
 modes. Hopefully most of the time will
 be in what's called Connected mode,

0:09:05.020000 --> 0:09:08.360000
 which means it is connected
 to the controller.

0:09:08.360000 --> 0:09:14.100000
 It's got a cap-wap data and a cap-wap control
 tunnel back to that controller,

0:09:14.100000 --> 0:09:16.500000
 or it could potentially operate
 in standalone mode.

0:09:16.500000 --> 0:09:19.640000
 This is sort of more of a disaster recovery
 type of thing, which means,

0:09:19.640000 --> 0:09:22.060000
 Hey, I've lost connectivity
 to the remote controller.

0:09:22.060000 --> 0:09:24.360000
 I'm doing standalone mode.

0:09:24.360000 --> 0:09:31.720000
 Now, let me go back here first of all
 to the second to the third and fourth

0:09:31.720000 --> 0:09:35.540000
 bullet points. Give myself
 a little space right here.

0:09:35.540000 --> 0:09:42.880000
 So if this is my branch, my branch office,
 and here's my lightweight access

0:09:42.880000 --> 0:09:48.160000
 point right here, and we've got
 perhaps an internet connection.

0:09:48.160000 --> 0:09:51.420000
 Let's draw that out.

0:09:51.420000 --> 0:09:55.780000
 It's connected to the internet
 back to headquarters.

0:09:55.780000 --> 0:09:58.680000
 And here is my controller.

0:09:58.680000 --> 0:10:04.520000
 Right there. There's
 my controller.

0:10:04.520000 --> 0:10:09.840000
 Now, we know that that lightweight access
 point is going to have to build

0:10:09.840000 --> 0:10:13.440000
 two tunnels back to
 that controller.

0:10:13.440000 --> 0:10:19.980000
 All right, what we call a control cap
-wap tunnel, which is built across

0:10:19.980000 --> 0:10:23.600000
 UDP, and the data.

0:10:23.600000 --> 0:10:29.860000
 Cap-wap tunnel, which is also built
 across UDP, just a different port

0:10:29.860000 --> 0:10:36.280000
 number. Now, if we have someone sitting
 right here and they're connected

0:10:36.280000 --> 0:10:42.300000
 to the lightweight access point, a logical
 question would be, all right,

0:10:42.300000 --> 0:10:46.300000
 how does the lightweight access point
 make the decision as far as when

0:10:46.300000 --> 0:10:51.060000
 this person sends a Wi-Fi frame, whether
 that Wi-Fi frame should go across

0:10:51.060000 --> 0:10:55.780000
 this data cap-wap tunnel to the controller
 and then be switched onto the

0:10:55.780000 --> 0:11:00.700000
 wired LAN from here, or should the lightweight
 access point actually switch

0:11:00.700000 --> 0:11:05.060000
 it itself onto a local LAN
 that is connected to?

0:11:05.060000 --> 0:11:09.540000
 Initially, you might think, oh, that
 must be based on the IP address of

0:11:09.540000 --> 0:11:12.020000
 where the data is going to.

0:11:12.020000 --> 0:11:15.720000
 Well, the lightweight access
 point is not a router.

0:11:15.720000 --> 0:11:19.400000
 It doesn't have the ability to look
 into the destination IP address of

0:11:19.400000 --> 0:11:22.160000
 the packets that that
 client is sending.

0:11:22.160000 --> 0:11:23.080000
 It can't do that.

0:11:23.080000 --> 0:11:25.880000
 This is a layer one and
 layer two device.

0:11:25.880000 --> 0:11:30.200000
 So the way that switching decision is
 made is not as dynamic as you might

0:11:30.200000 --> 0:11:36.540000
 think. It's all based on what SSID
 that client is connected to.

0:11:36.540000 --> 0:11:40.360000
 So on the controller, when the controller
 says, hey, access point, you

0:11:40.360000 --> 0:11:41.920000
 can operate in flex-connect mode.

0:11:41.920000 --> 0:11:46.580000
 The controller will be configured on
 the controller's GUI to push down

0:11:46.580000 --> 0:11:51.680000
 at least a couple of SSIDs
 down to that lap.

0:11:51.680000 --> 0:11:55.380000
 When you define those SSIDs on the
 controller, you will actually tell

0:11:55.380000 --> 0:12:00.220000
 the controller, hey, this SSID right
 here, if someone associates to that

0:12:00.220000 --> 0:12:05.140000
 on the lap, that means their data traffic
 is going to be switched on the

0:12:05.140000 --> 0:12:09.940000
 lap itself. The lap itself will take
 their Wi-Fi frames, convert them

0:12:09.940000 --> 0:12:14.060000
 into wired Ethernet frames, and dump
 them right out natively as Ethernet

0:12:14.060000 --> 0:12:16.120000
 onto the wired LAN.

0:12:16.120000 --> 0:12:18.280000
 So that SSID will be used.

0:12:18.280000 --> 0:12:21.320000
 You connect to that when you want to
 connect to the branch office stuff

0:12:21.320000 --> 0:12:24.680000
 like a printer at the branch office,
 or maybe a file server at the branch

0:12:24.680000 --> 0:12:28.900000
 office. Then there will be another
 SSID that's created, and that one,

0:12:28.900000 --> 0:12:32.440000
 the controller will say, hey, if somebody
 associates to that, they have

0:12:32.440000 --> 0:12:36.720000
 to go across the CAPWAP data tunnel,
 all their data will go across that.

0:12:36.720000 --> 0:12:40.420000
 So it's not based on the IP address of
 the actual packets from the client,

0:12:40.420000 --> 0:12:44.560000
 it's based on whichever SSID
 they actually connect to.

0:12:44.560000 --> 0:12:50.480000
 That's how the switching decision
 is made by FlexConnect.

0:12:50.480000 --> 0:12:56.820000
 Now, so all of this has been talking
 about central centralized wireless

0:12:56.820000 --> 0:13:01.300000
 architectures, and FlexConnect was a
 way to help alleviate some of those

0:13:01.300000 --> 0:13:05.460000
 hairpins, depending on which
 SSID you're connected to.

0:13:05.460000 --> 0:13:11.500000
 Another possibility, another type of wireless
 architecture is called converged

0:13:11.500000 --> 0:13:12.700000
 wireless network architecture.

0:13:12.700000 --> 0:13:17.760000
 And the idea here is, hey, instead
 of having our controller locate an

0:13:17.760000 --> 0:13:23.120000
 essential spot like our data center,
 where there's long CAPWAP tunnels

0:13:23.120000 --> 0:13:26.700000
 that have to be built to reach
 that guy, why don't we do this?

0:13:26.700000 --> 0:13:30.420000
 Why don't we take that functionality
 of the controller and move it down

0:13:30.420000 --> 0:13:33.860000
 to like the distribution
 or the access layer?

0:13:33.860000 --> 0:13:38.720000
 As a matter of fact, why don't we take
 the switch, the layer, the land

0:13:38.720000 --> 0:13:41.940000
 switch to that access point is physically
 connected to on his ethernet

0:13:41.940000 --> 0:13:46.380000
 port, and why don't we put the controller
 functionality right there in

0:13:46.380000 --> 0:13:50.300000
 the switch? Now, there's still going
 to be two CAPWAP tunnels built, a

0:13:50.300000 --> 0:13:53.320000
 data tunnel and a control tunnel, but
 now they're going to be really,

0:13:53.320000 --> 0:13:57.080000
 really short, because they're only
 going from the access point, maybe

0:13:57.080000 --> 0:14:02.980000
 5, 10, 20 feet to the switch it's
 connected to on the other side.

0:14:02.980000 --> 0:14:06.760000
 This is called a converged wireless
 network architecture, where we're

0:14:06.760000 --> 0:14:10.900000
 basically implementing multiple controllers
 to divide the load from the

0:14:10.900000 --> 0:14:17.100000
 access points. Now, historically, this
 used to be done by actually taking

0:14:17.100000 --> 0:14:22.320000
 Wi-Fi controller modules, physical modules
 you can hold, and sliding them

0:14:22.320000 --> 0:14:25.480000
 into an available slot
 in a switch's chassis.

0:14:25.480000 --> 0:14:29.120000
 So you had to have a switch that was
 a module or a switch like a 6500

0:14:29.120000 --> 0:14:33.860000
 or a 6800 or something, by one of these
 special controller modules, slide

0:14:33.860000 --> 0:14:37.580000
 it there into the chassis, and that was
 called converged wireless, because

0:14:37.580000 --> 0:14:41.280000
 now the switch was doing switching,
 and it had the special module that

0:14:41.280000 --> 0:14:44.800000
 could act as a wireless
 LAN controller as well.

0:14:44.800000 --> 0:14:48.460000
 Cisco is sort of moving away from that
 now, and now what they're doing

0:14:48.460000 --> 0:14:51.840000
 is they're saying, hey, if you buy
 the special Cisco switch called the

0:14:51.840000 --> 0:14:55.460000
 Cisco Catalyst 9300, and I'm sure they're
 going to create some more in

0:14:55.460000 --> 0:15:01.480000
 the future as well, well, this 9300 is
 running iOS XE software, the latest

0:15:01.480000 --> 0:15:05.700000
 and greatest software for programmability
 and network automations, got

0:15:05.700000 --> 0:15:10.220000
 all sorts of APIs and stuff built into
 it, and hey, guess what one of

0:15:10.220000 --> 0:15:12.900000
 the features is that
 iOS XE gives you?

0:15:12.900000 --> 0:15:17.800000
 If you have it in a 9300 switch, you
 now also have the ability to turn

0:15:17.800000 --> 0:15:22.260000
 on wireless LAN controller functionality
 right there within the switch

0:15:22.260000 --> 0:15:28.820000
 itself. So the idea here is very simple,
 that your lap, your light weight

0:15:28.820000 --> 0:15:32.560000
 access point would be physically connected
 to that switch, and in that

0:15:32.560000 --> 0:15:36.140000
 exact same switch, it's running
 the controller functionality.

0:15:36.140000 --> 0:15:39.300000
 So here you can see the CAPWAP data
 and control tunnels now are very,

0:15:39.300000 --> 0:15:45.060000
 very short. We don't even really have
 to worry about latency at this point.

0:15:45.060000 --> 0:15:49.620000
 So there's some many benefits and some
 drawbacks with the converged network

0:15:49.620000 --> 0:15:51.260000
 architecture as well.

0:15:51.260000 --> 0:15:54.640000
 So the benefits are you've got a
 very short CAPWAP data tunnel.

0:15:54.640000 --> 0:15:58.320000
 This reduces the impact of hairpins,
 we don't have to worry about really

0:15:58.320000 --> 0:16:00.980000
 long round trip transmit times.

0:16:00.980000 --> 0:16:06.160000
 This is also better support for
 faster Wi-Fi technologies.

0:16:06.160000 --> 0:16:10.900000
 You see, up until the last few years, your
 Wi-Fi clients, when they connected

0:16:10.900000 --> 0:16:14.600000
 to the access point, the actual speeds
 at which they transmitted those

0:16:14.600000 --> 0:16:17.560000
 clients was under 1 gigabit.

0:16:17.560000 --> 0:16:20.400000
 Typically speaking, it was like in the
 hundreds, if you were talking about

0:16:20.400000 --> 0:16:25.340000
 802.11N, you were lucky to get
 like 300 megabits per second.

0:16:25.340000 --> 0:16:26.700000
 Lucky to get that.

0:16:26.700000 --> 0:16:33.100000
 Well, with some of the later Wi-Fi
 developments like 802.11AC phase 2,

0:16:33.100000 --> 0:16:39.080000
 802.11AD and the up and coming 802
.11AX, now we have a situation where

0:16:39.080000 --> 0:16:43.520000
 a Wi-Fi client, if it's correctly positioned
 and really close to the access

0:16:43.520000 --> 0:16:48.100000
 point, if the stars and the moon align
 just perfectly, that Wi-Fi client

0:16:48.100000 --> 0:16:51.960000
 could actually be transmitting
 at faster than gigabit speeds.

0:16:51.960000 --> 0:16:55.620000
 It could be transmitted at 1.5 gigs,
 two gigs, or maybe even faster than

0:16:55.620000 --> 0:17:01.120000
 that. Well, most access points, if you're
 going to transmit it that fast,

0:17:01.120000 --> 0:17:05.660000
 first of all, number one, most access points
 have a 10, 100, 1,000 connection

0:17:05.660000 --> 0:17:07.000000
 on the back of them.

0:17:07.000000 --> 0:17:10.400000
 How's that access point going to handle
 incoming traffic from the Wi-Fi

0:17:10.400000 --> 0:17:14.820000
 client? It's coming in at like 1.5 gigabits,
 when the maximum it can transmit

0:17:14.820000 --> 0:17:18.320000
 on the outside to the
 switch, is 1 gigabit.

0:17:18.320000 --> 0:17:19.860000
 Well, here's a good thing.

0:17:19.860000 --> 0:17:22.680000
 First of all, you wouldn't buy an access
 point that only has one interface

0:17:22.680000 --> 0:17:25.180000
 like that. You'd be setting
 yourself up for failure.

0:17:25.180000 --> 0:17:30.380000
 So you'd buy a higher end access point
 that has two of those 10, 100,

0:17:30.380000 --> 0:17:34.480000
 1,000 interfaces, and then you could
 bundle them together into a link

0:17:34.480000 --> 0:17:35.860000
 aggregation group.

0:17:35.860000 --> 0:17:38.520000
 We're not going to talk about that, but
 if you're familiar with switching,

0:17:38.520000 --> 0:17:41.980000
 basically the same thing
 is ether channeling.

0:17:41.980000 --> 0:17:45.560000
 Now you've got your bandwidth that
 can handle those really fast Wi-Fi

0:17:45.560000 --> 0:17:47.180000
 clients. Let's say
 you've done that.

0:17:47.180000 --> 0:17:48.700000
 You already finished with that.

0:17:48.700000 --> 0:17:49.960000
 But now think about this.

0:17:49.960000 --> 0:17:54.820000
 Those clients are transmitting
 at 1.5 gigabits or faster.

0:17:54.820000 --> 0:17:57.740000
 Through a capwap data tunnel.

0:17:57.740000 --> 0:18:01.020000
 What if that data tunnel terminates on
 a central wireless controller that's

0:18:01.020000 --> 0:18:03.580000
 like six router hops away?

0:18:03.580000 --> 0:18:07.660000
 Now we might have a problem with getting
 that 1.5 or two or three gigabits

0:18:07.660000 --> 0:18:12.840000
 of traffic up to that controller and
 not delaying it, not dropping it.

0:18:12.840000 --> 0:18:15.760000
 Well, with converged wireless, we don't
 have to worry about that because

0:18:15.760000 --> 0:18:20.280000
 the tunnel ends right there at the connected
 switch that the access point

0:18:20.280000 --> 0:18:25.440000
 is connected to.

0:18:25.440000 --> 0:18:29.440000
 Also, we have less access points
 per embedded controller.

0:18:29.440000 --> 0:18:32.740000
 Sometimes we call these embedded controllers
 because the controller functionality

0:18:32.740000 --> 0:18:37.100000
 is embedded right into the
 local switch itself.

0:18:37.100000 --> 0:18:41.660000
 So what we're talking about here is in
 a centralized wireless architecture,

0:18:41.660000 --> 0:18:46.200000
 a controller might actually be controlling
 potentially thousands of access

0:18:46.200000 --> 0:18:48.960000
 points, just depending on
 how big the situation is.

0:18:48.960000 --> 0:18:53.220000
 Here, in a converged wireless architecture,
 where it's built into the

0:18:53.220000 --> 0:18:57.100000
 switch, that switch is only going to
 be controlling access points are

0:18:57.100000 --> 0:19:00.700000
 probably physically connected
 on its switch ports.

0:19:00.700000 --> 0:19:04.940000
 So we might be talking about maybe a few
 dozen access points at the absolute

0:19:04.940000 --> 0:19:10.120000
 most. So it's a lot less taxing
 for the controller to do that.

0:19:10.120000 --> 0:19:14.500000
 And now we can have controller functionality
 right there at the branch

0:19:14.500000 --> 0:19:18.560000
 site. Now, there are just a couple,
 just really sort of one big drawback

0:19:18.560000 --> 0:19:20.960000
 to this, which is dollars.

0:19:20.960000 --> 0:19:25.300000
 Okay? So if you're dealing with converged
 wireless architecture where

0:19:25.300000 --> 0:19:28.520000
 you're actually buying some of the older
 modules, the physical controllers

0:19:28.520000 --> 0:19:32.780000
 that slip into a switch, well,
 you got to buy those things.

0:19:32.780000 --> 0:19:34.360000
 Those things aren't free.

0:19:34.360000 --> 0:19:35.840000
 They're kind of pricey.

0:19:35.840000 --> 0:19:39.080000
 And so if you think about this now,
 now you think about, okay, all of

0:19:39.080000 --> 0:19:43.020000
 our access layer switches have to
 have these modules put into them.

0:19:43.020000 --> 0:19:46.980000
 To support the access points are physically
 connected to those switches.

0:19:46.980000 --> 0:19:50.620000
 That's a little bit more pricey than
 buying just one controller appliance

0:19:50.620000 --> 0:19:53.340000
 which is sitting up in the
 data center somewhere.

0:19:53.340000 --> 0:19:57.580000
 Or you say, well, hey, I'm getting
 the 9300 series switch, so I don't

0:19:57.580000 --> 0:19:59.220000
 have to buy a separate controller.


0:19:59.220000 --> 0:20:04.440000
 Okay, great. But that 9300 series switch
 is not the cheapest switch that

0:20:04.440000 --> 0:20:07.440000
 Cisco sells. It's actually
 a fairly pricey switch.

0:20:07.440000 --> 0:20:10.380000
 It's on the higher end as far as
 dollar amounts are concerned.

0:20:10.380000 --> 0:20:14.440000
 So either way, if you want to go with
 a converged network architecture

0:20:14.440000 --> 0:20:18.580000
 for wireless, it's going to cost you
 a little bit more money to do this.

0:20:18.580000 --> 0:20:21.580000
 But as you can see, there are
 some great benefits to this.

0:20:21.580000 --> 0:20:24.840000
 So that concludes this video.

0:20:24.840000 --> 0:20:25.720000
 Thank you for watching.

