WEBVTT

0:00:02.800000 --> 0:00:08.220000
 Hello and welcome to this video titled
 Network Topology Architectures.

0:00:08.220000 --> 0:00:12.400000
 In this video we're going to cover
 several different types of network

0:00:12.400000 --> 0:00:15.840000
 architectures. I'm going to introduce
 you to some concepts such as two

0:00:15.840000 --> 0:00:18.940000
-tier versus three-tier
 architectures.

0:00:18.940000 --> 0:00:21.420000
 What is a spine leaf architecture?


0:00:21.420000 --> 0:00:24.700000
 What type of WAN architectures
 exist?

0:00:24.700000 --> 0:00:29.380000
 What about small office home office
 architectures and on premise versus

0:00:29.380000 --> 0:00:31.560000
 cloud-based architectures?

0:00:31.560000 --> 0:00:34.040000
 I'm going to introduce you to the differences
 in all these and what some

0:00:34.040000 --> 0:00:37.340000
 of the characteristics are
 of each one of these.

0:00:37.340000 --> 0:00:42.600000
 So let's start by answering the question,
 what is meant by a network architecture?

0:00:42.600000 --> 0:00:47.500000
 What is that? Well, networks clearly,
 as you can probably guess by this

0:00:47.500000 --> 0:00:50.480000
 point, can be designed
 in a variety of ways.

0:00:50.480000 --> 0:00:54.620000
 Not only do you get to choose what
 devices you need for your networks,

0:00:54.620000 --> 0:00:58.260000
 do I need a switch, do I need a router,
 do I need a wireless access point,

0:00:58.260000 --> 0:01:02.300000
 do I need a controller, how many of
 these things do I need, but you also

0:01:02.300000 --> 0:01:06.200000
 get to choose how these things are interconnected
 and where within your

0:01:06.200000 --> 0:01:09.580000
 company or building are they going
 to be physically located.

0:01:09.580000 --> 0:01:13.760000
 So all of those things will determine
 what type of network architecture

0:01:13.760000 --> 0:01:19.660000
 you use. So depending on your positioning
 and placement of all these devices,

0:01:19.660000 --> 0:01:23.360000
 it will dictate things such as what
 types of cables you need and where

0:01:23.360000 --> 0:01:26.340000
 you need to place those cables
 and how long they need to be.

0:01:26.340000 --> 0:01:28.660000
 Where will the data go?

0:01:28.660000 --> 0:01:33.080000
 So when John's laptop is talking to
 the server that hosts all the files,

0:01:33.080000 --> 0:01:35.620000
 what will the path
 of that data be?

0:01:35.620000 --> 0:01:39.460000
 And if that path goes down, is there
 an alternate or redundant path that

0:01:39.460000 --> 0:01:43.180000
 can be used? It will also
 talk about redundancy.

0:01:43.180000 --> 0:01:46.220000
 All these different architectures somehow
 have more or less redundancy

0:01:46.220000 --> 0:01:50.280000
 than others. And in case you're not
 familiar with that term, what that's

0:01:50.280000 --> 0:01:54.720000
 simply referring to is if a path in a
 network goes down like a cable gets

0:01:54.720000 --> 0:01:58.340000
 yanked out, someone accidentally spills
 coffee in a router or switch and

0:01:58.340000 --> 0:02:03.200000
 it's short circuits, I want some alternative
 path available from my data

0:02:03.200000 --> 0:02:05.280000
 to still reach its destination.

0:02:05.280000 --> 0:02:06.500000
 That's redundancy.

0:02:06.500000 --> 0:02:11.180000
 The more alternative paths you have,
 the more redundancy you have.

0:02:11.180000 --> 0:02:14.860000
 Also, some of these architectures will
 dictate, do I need to use third

0:02:14.860000 --> 0:02:18.980000
-party services? In other words, can
 I make the architecture all on my

0:02:18.980000 --> 0:02:23.800000
 own, all with my own cabling and
 my own networking devices?

0:02:23.800000 --> 0:02:27.940000
 Or do I need to contract the services,
 some outside provider to help me

0:02:27.940000 --> 0:02:32.320000
 build some or most of the architecture
 I want to use?

0:02:32.320000 --> 0:02:35.800000
 So, let's start by looking at some of
 these architectures, starting with

0:02:35.800000 --> 0:02:39.400000
 two tier versus three
 tier architectures.

0:02:39.400000 --> 0:02:44.340000
 Now, these architectures here typically
 where you will find them is within

0:02:44.340000 --> 0:02:47.160000
 your building. So, we're talking about
 architectures here that are not

0:02:47.160000 --> 0:02:51.040000
 for your house, but
 for your company.

0:02:51.040000 --> 0:02:55.920000
 So, a company's network and these are
 typically mid to larger size companies

0:02:55.920000 --> 0:02:58.880000
 that will implement these type of architectures
 I'm going to show you

0:02:58.880000 --> 0:03:02.780000
 right here. So, these are implemented
 in your local area network, in your

0:03:02.780000 --> 0:03:05.580000
 LAN. Let's start with three tier.

0:03:05.580000 --> 0:03:08.420000
 Historically, three
 tier first came out.

0:03:08.420000 --> 0:03:12.940000
 This was the first design methodology
 for local area networks.

0:03:12.940000 --> 0:03:14.620000
 And here's how it worked.

0:03:14.620000 --> 0:03:18.540000
 The idea was that we're going to take
 our network and divide up into three

0:03:18.540000 --> 0:03:19.840000
 different tiers.

0:03:19.840000 --> 0:03:23.660000
 Each tier is going to have a different
 name and a different responsibility.

0:03:23.660000 --> 0:03:26.900000
 So, the lowest tier is what
 we call the axis layer.

0:03:26.900000 --> 0:03:32.320000
 These are the network devices that are
 giving my hosts such as my laptops,

0:03:32.320000 --> 0:03:38.460000
 my PCs, my servers, my tablets, physical
 connectivity to the network.

0:03:38.460000 --> 0:03:42.860000
 So, whatever the very first networking
 device is that my laptop, tablet

0:03:42.860000 --> 0:03:48.160000
 or PC touches, that device
 is at the axis layer.

0:03:48.160000 --> 0:03:50.580000
 Typically, that's going
 to be one of two things.

0:03:50.580000 --> 0:03:55.760000
 It's either going to be network based
 switches or it's going to be wireless

0:03:55.760000 --> 0:04:00.300000
 access. That's what you would typically
 see at the axis layer.

0:04:00.300000 --> 0:04:05.980000
 So, the primary job of the axis layer
 is, like it sounds, to provide network

0:04:05.980000 --> 0:04:10.820000
 access to whatever devices want to connect,
 and also, possibly to implement

0:04:10.820000 --> 0:04:15.480000
 authentication. Make sure that unauthorized
 devices are not allowed to

0:04:15.480000 --> 0:04:18.020000
 connect and gain access
 to our network.

0:04:18.020000 --> 0:04:24.080000
 Now, after that, the next layer up
 is called the distribution layer.

0:04:24.080000 --> 0:04:28.500000
 This would also be composed of either
 routers or switches that can do

0:04:28.500000 --> 0:04:34.160000
 routing. The main idea here is that
 the axis layer, we may have grouped

0:04:34.160000 --> 0:04:38.060000
 together our hosts into different groups,
 such as maybe this group of

0:04:38.060000 --> 0:04:40.300000
 hosts here is in the
 HR department.

0:04:40.300000 --> 0:04:46.960000
 This guy over here
 is in marketing.

0:04:46.960000 --> 0:04:53.340000
 If HR wants to send data to this marketing
 server, the access layer devices

0:04:53.340000 --> 0:04:55.480000
 might not know what that
 grouping looks like.

0:04:55.480000 --> 0:04:59.260000
 They might not know how to get data
 from one group to the other, but the

0:04:59.260000 --> 0:05:01.440000
 distribution layer will.

0:05:01.440000 --> 0:05:05.900000
 So, data that needs to go from one
 segment or one group of the network

0:05:05.900000 --> 0:05:11.120000
 to another will pass through the distribution
 layer and then go back down

0:05:11.120000 --> 0:05:15.480000
 again. So, the distribution layer
 is in charge of routing.

0:05:15.480000 --> 0:05:19.180000
 It might also be in
 charge of security.

0:05:19.180000 --> 0:05:24.180000
 For example, if we say, okay, well,
 HR is allowed to go to marketing,

0:05:24.180000 --> 0:05:27.860000
 but HR is not allowed to
 send data to payroll.

0:05:27.860000 --> 0:05:32.600000
 Well, whatever networking security features
 we implement would be implemented

0:05:32.600000 --> 0:05:36.340000
 at the distribution layer to
 contain that or control that.

0:05:36.340000 --> 0:05:39.780000
 And then lastly, in the three layer
 architecture, we have what's called

0:05:39.780000 --> 0:05:44.660000
 the core layer. The idea behind the
 core layer is that this is a super,

0:05:44.660000 --> 0:05:49.380000
 super fast layer that's meant just to
 get data from like major sections

0:05:49.380000 --> 0:05:52.700000
 of our network to other major
 sections of our network.

0:05:52.700000 --> 0:05:59.040000
 So, for example, you might have two
 buildings, two separate buildings.

0:05:59.040000 --> 0:06:02.680000
 Each building has a bunch of access layer
 devices like switches and wireless

0:06:02.680000 --> 0:06:07.360000
 access points. Each building has some
 stuff at the distribution layer.

0:06:07.360000 --> 0:06:11.340000
 So, in building A, any information
 that has to be routed or sent from

0:06:11.340000 --> 0:06:15.560000
 corporate payroll to corporate marketing,
 both of which is in building

0:06:15.560000 --> 0:06:19.020000
 A, would just stay in the distribution
 and access layer.

0:06:19.020000 --> 0:06:20.460000
 It would just stay there.

0:06:20.460000 --> 0:06:25.020000
 However, if anybody in building A needs
 to send data to another group

0:06:25.020000 --> 0:06:28.960000
 that's in building B, now we send it
 all the way to the core layer and

0:06:28.960000 --> 0:06:32.800000
 the core layer is responsible for super
 fast switching of data from one

0:06:32.800000 --> 0:06:33.500000
 major section of data.

0:06:33.500000 --> 0:06:36.200000
 So, in the other section of the network,
 like a building to another major

0:06:36.200000 --> 0:06:38.900000
 section of the network,
 like another building.

0:06:38.900000 --> 0:06:43.640000
 So, the data should not be slowed
 down at all in the core layer.

0:06:43.640000 --> 0:06:47.500000
 The idea here behind this three layer
 or three tier model is that there's

0:06:47.500000 --> 0:06:49.200000
 going to be any delay in the data.


0:06:49.200000 --> 0:06:51.700000
 If there's anything that's going to
 be slowing it down, it's going to

0:06:51.700000 --> 0:06:54.280000
 happen right here at the
 distribution layer.

0:06:54.280000 --> 0:06:57.600000
 Any routing or security features or
 anything that could potentially slow

0:06:57.600000 --> 0:07:01.300000
 the data down where we have to look
 at the data, inspect the data, make

0:07:01.300000 --> 0:07:05.500000
 a decision on the data, would happen right
 there at the distribution layer.

0:07:05.500000 --> 0:07:09.060000
 If data makes its way all the way up
 to the core layer, these things here

0:07:09.060000 --> 0:07:12.700000
 are super, super fast
 networking devices.

0:07:12.700000 --> 0:07:16.500000
 Their job is just to whip stuff across
 really quickly so that it can go

0:07:16.500000 --> 0:07:19.320000
 down again. That's the core layer.


0:07:19.320000 --> 0:07:23.620000
 Now, this three layer model, this is
 how Cisco recommended that networks

0:07:23.620000 --> 0:07:26.420000
 be built for a long, long time.

0:07:26.420000 --> 0:07:32.360000
 And this is because 20, 30 years ago,
 there were very distinct, there

0:07:32.360000 --> 0:07:33.680000
 were different devices.

0:07:33.680000 --> 0:07:36.660000
 You actually had to buy a different
 type of device to do a core layer

0:07:36.660000 --> 0:07:42.140000
 responsibility versus a distribution
 layer responsibility versus an access

0:07:42.140000 --> 0:07:43.180000
 layer responsibility.

0:07:43.180000 --> 0:07:45.480000
 You had to buy different
 devices to do this.

0:07:45.480000 --> 0:07:49.860000
 Well, over time, more and more functionality
 has been built into devices

0:07:49.860000 --> 0:07:54.200000
 to where these days, customers say,
 you know what, I don't need to buy

0:07:54.200000 --> 0:07:59.480000
 all this stuff to do the same types
 of things, the types of operations

0:07:59.480000 --> 0:08:03.560000
 that are done at core,
 distribution or access.

0:08:03.560000 --> 0:08:09.000000
 I can actually shrink this down to less
 layers and accomplish the exact

0:08:09.000000 --> 0:08:13.940000
 same thing. And that brings us over
 to the two tier architecture.

0:08:13.940000 --> 0:08:19.300000
 This is also sometimes called the
 collapsed core architecture.

0:08:19.300000 --> 0:08:23.480000
 We don't really necessarily have a
 name for each layer at this point.

0:08:23.480000 --> 0:08:27.060000
 I guess technically you could still
 say this could still be referred to

0:08:27.060000 --> 0:08:31.740000
 as the access layer, possibly because
 after all, it is giving access to

0:08:31.740000 --> 0:08:33.880000
 the network to our hosts.

0:08:33.880000 --> 0:08:38.120000
 But the devices right here in layer
 two, these are actually doing what

0:08:38.120000 --> 0:08:39.400000
 this used to do.

0:08:39.400000 --> 0:08:43.860000
 So my distribution and core layer are
 now being handled in just one tier

0:08:43.860000 --> 0:08:50.140000
 instead of two. Hence, this is why they
 call it a two tier architecture.

0:08:50.140000 --> 0:08:54.580000
 Now, what are some of the characteristics,
 regardless of what hardware

0:08:54.580000 --> 0:08:58.680000
 you buy or what software you use, what
 are some of the common characteristics

0:08:58.680000 --> 0:09:02.200000
 of two tier or three
 tier architectures?

0:09:02.200000 --> 0:09:04.460000
 Well, there's three things
 I could think of.

0:09:04.460000 --> 0:09:10.940000
 Number one, this type of architecture
 allows for lots of different types

0:09:10.940000 --> 0:09:13.260000
 of things at the access layer.

0:09:13.260000 --> 0:09:18.100000
 Because host devices have a variety of
 ways of connecting to the network.

0:09:18.100000 --> 0:09:21.680000
 They might use wired ethernet, in which
 case a network based switch would

0:09:21.680000 --> 0:09:26.060000
 be appropriate. They might use Wi-Fi,
 in which an access layer switches

0:09:26.060000 --> 0:09:29.660000
 appropriate. Hey, they might even use
 Bluetooth or something else, in

0:09:29.660000 --> 0:09:32.020000
 which another device
 might be appropriate.

0:09:32.020000 --> 0:09:36.080000
 So the access layer utilizes several
 different types of devices, or I

0:09:36.080000 --> 0:09:39.460000
 should say could utilize several
 different types of devices.

0:09:39.460000 --> 0:09:44.820000
 In order to accommodate a wide
 variety of host devices.

0:09:44.820000 --> 0:09:46.440000
 Okay, what else?

0:09:46.440000 --> 0:09:49.220000
 What else is a common characteristic
 of this?

0:09:49.220000 --> 0:09:54.760000
 Well, a lot of times you will find
 security implemented here.

0:09:54.760000 --> 0:09:58.820000
 At the access layer we might have security
 protocols that differentiate

0:09:58.820000 --> 0:10:02.020000
 who's allowed onto the
 network and who's not.

0:10:02.020000 --> 0:10:04.560000
 That would be a very big concern
 of these networks.

0:10:04.560000 --> 0:10:09.580000
 So in your LANs, a lot of times authentication
 is done on your access

0:10:09.580000 --> 0:10:14.800000
 layer devices. And also another characteristic
 of this is we want quick

0:10:14.800000 --> 0:10:18.520000
 and easy connection
 to these devices.

0:10:18.520000 --> 0:10:21.400000
 So we want to provide
 very common ports.

0:10:21.400000 --> 0:10:25.200000
 We want to make the users connection
 to the network very easy so they

0:10:25.200000 --> 0:10:27.960000
 don't have to jump through a lot
 of hoops to get onto the network.

0:10:27.960000 --> 0:10:32.580000
 So however that's done to make it easy
 and streamlined would be done at

0:10:32.580000 --> 0:10:34.040000
 the access layer.

0:10:34.040000 --> 0:10:41.200000
 Alright, so this is for your campus or
 your enterprise network where your

0:10:41.200000 --> 0:10:42.760000
 hosts are going to connect.

0:10:42.760000 --> 0:10:47.320000
 Your laptops, your PCs, your wireless
 devices like your cell phone, your

0:10:47.320000 --> 0:10:50.960000
 smart phones and your tablets.

0:10:50.960000 --> 0:10:55.920000
 But we also know that a network has another
 section where all of our file

0:10:55.920000 --> 0:10:59.260000
 servers live. And this is what
 we call our data center.

0:10:59.260000 --> 0:11:01.920000
 Our data centers where all our file
 servers live that have all of our

0:11:01.920000 --> 0:11:03.460000
 mission critical data.

0:11:03.460000 --> 0:11:06.620000
 So you can sort of think of your network
 within your company as subdivided

0:11:06.620000 --> 0:11:09.040000
 into two main categories.

0:11:09.040000 --> 0:11:13.140000
 You've got the campus or enterprise
 LAN which is all of your hosts live

0:11:13.140000 --> 0:11:16.620000
 and the networks used to get
 data back and forth to them.

0:11:16.620000 --> 0:11:20.520000
 But most of the time they're trying
 to access data in your servers.

0:11:20.520000 --> 0:11:24.700000
 And your servers are all over here
 in a much more secure part of your

0:11:24.700000 --> 0:11:27.180000
 network called the data center.

0:11:27.180000 --> 0:11:30.780000
 Well the architecture for a data center
 is not necessarily going to use

0:11:30.780000 --> 0:11:34.420000
 this. A data center will use a different
 type of architecture which we

0:11:34.420000 --> 0:11:37.300000
 call a spine leaf architecture.

0:11:37.300000 --> 0:11:43.900000
 So when you hear this term spine leaf
 architecture think data center.

0:11:43.900000 --> 0:11:48.800000
 Okay so in a spine leaf architecture
 interconnections between switches

0:11:48.800000 --> 0:11:51.140000
 can be layer two or layer three.

0:11:51.140000 --> 0:11:53.920000
 Alright so let's dig into
 this a little bit more.

0:11:53.920000 --> 0:12:00.660000
 First of all what makes this different
 than what we just saw?

0:12:00.660000 --> 0:12:07.360000
 Well number one while of course our
 ultimate objective is that no part

0:12:07.360000 --> 0:12:09.140000
 of the network should
 ever go down.

0:12:09.140000 --> 0:12:11.820000
 We don't want any part of the
 network to ever go down.

0:12:11.820000 --> 0:12:15.240000
 But if you think about it if I'm dividing
 my network into the data center

0:12:15.240000 --> 0:12:19.500000
 which has my mission critical stuff, my
 data that's stored here and everything

0:12:19.500000 --> 0:12:24.560000
 else. If I have to decide between those
 two my data center is much more

0:12:24.560000 --> 0:12:28.880000
 valuable to me. That really
 can't ever go down.

0:12:28.880000 --> 0:12:31.420000
 Yeah I don't want this section over
 the network over here that's holding

0:12:31.420000 --> 0:12:33.480000
 my hosts. I don't want
 that to go down either.

0:12:33.480000 --> 0:12:37.700000
 But if I have to pick between the two
 I'll spend the bigger bucks, get

0:12:37.700000 --> 0:12:42.820000
 the better faster more resilient equipment
 and put it over here in the

0:12:42.820000 --> 0:12:47.240000
 data center. So one characteristic here
 of spine leaf is that these switches

0:12:47.240000 --> 0:12:51.700000
 that you see, these aren't the same
 switches that your laptops and PCs

0:12:51.700000 --> 0:12:53.080000
 are connecting to.

0:12:53.080000 --> 0:12:56.900000
 Those are pretty pricey,
 these are super pricey.

0:12:56.900000 --> 0:13:01.040000
 So the switches here in the spine leaf
 architecture are designed with

0:13:01.040000 --> 0:13:06.280000
 the intent and mind of number one providing
 maximum resiliency and redundancy.

0:13:06.280000 --> 0:13:09.540000
 These things are advertised
 as these will never go down.

0:13:09.540000 --> 0:13:13.020000
 You buy one of these things, you can
 walk away, never have to worry about

0:13:13.020000 --> 0:13:15.700000
 it going down. Of course you're going
 to pay a premium for that because

0:13:15.700000 --> 0:13:19.720000
 their power supplies are going to be
 super expensive to support that.

0:13:19.720000 --> 0:13:23.300000
 They'll probably have multiple power supplies
 in case one goes down, another

0:13:23.300000 --> 0:13:24.860000
 one can take up the job.

0:13:24.860000 --> 0:13:27.920000
 So they'll have a whole bunch of resiliency
 and redundancy features built

0:13:27.920000 --> 0:13:31.100000
 into them. So that's one thing that
 makes this architecture different.

0:13:31.100000 --> 0:13:34.760000
 Another thing is that because this
 is where our mission critical data

0:13:34.760000 --> 0:13:38.780000
 is, there's a lot of data
 coming to this spine leaf.

0:13:38.780000 --> 0:13:43.060000
 There's a lot of hosts spread throughout
 our company and all their requests,

0:13:43.060000 --> 0:13:46.240000
 all their data is funneling
 into this.

0:13:46.240000 --> 0:13:50.920000
 So these connections have
 to be super, super fast.

0:13:50.920000 --> 0:13:54.620000
 So that's another difference, is in a spine
 leaf architecture these interconnections

0:13:54.620000 --> 0:13:58.260000
 between the switches and the connections
 leading down to the servers down

0:13:58.260000 --> 0:14:02.640000
 below are typically going to be much
 faster connections than what you

0:14:02.640000 --> 0:14:06.080000
 would find in the two tier or three
 tier, which is also going to drive

0:14:06.080000 --> 0:14:10.240000
 up your price. The faster the connections,
 the more you pay for those

0:14:10.240000 --> 0:14:13.860000
 types of ports and interfaces
 on routers and switches.

0:14:13.860000 --> 0:14:19.840000
 So, and then lastly, notice multiple
 redundant connections here.

0:14:19.840000 --> 0:14:24.420000
 Every switch on the bottom has a connection
 to every switch on the top.

0:14:24.420000 --> 0:14:28.560000
 And the idea here is that this type
 of spine leaf architecture is best

0:14:28.560000 --> 0:14:32.740000
 suited for what they call
 East West traffic.

0:14:32.740000 --> 0:14:37.680000
 In other words, traffic that's going
 like this or going like this or going

0:14:37.680000 --> 0:14:40.460000
 like this. This is considered
 East West traffic.

0:14:40.460000 --> 0:14:43.900000
 The idea being that there's not going
 to be a lot of traffic going north

0:14:43.900000 --> 0:14:48.220000
 south. There will be some, but more
 the traffic is going back and forth

0:14:48.220000 --> 0:14:49.780000
 than up and down.

0:14:49.780000 --> 0:14:54.460000
 So that's why we have all these redundant
 connections here between our

0:14:54.460000 --> 0:14:59.260000
 spine switches. So these ones here are
 called our spine switches and these

0:14:59.260000 --> 0:15:02.840000
 here are called our leaf switches.


0:15:02.840000 --> 0:15:06.180000
 Hence where we get the
 term spine leaf.

0:15:06.180000 --> 0:15:11.080000
 One other thing about this architecture
 I want you to be aware of, there's

0:15:11.080000 --> 0:15:15.720000
 another term for this which
 is called the fabric.

0:15:15.720000 --> 0:15:18.740000
 So if you ever hear somebody saying,
 oh yeah, my data center, my fabric

0:15:18.740000 --> 0:15:21.280000
 is made up of blah, blah,
 blah, blah, blah.

0:15:21.280000 --> 0:15:25.360000
 Well, the fabric, that means they're
 in a spine leaf architecture.

0:15:25.360000 --> 0:15:30.560000
 And the fabric is all these switches,
 super fast, super redundant, super

0:15:30.560000 --> 0:15:34.320000
 expensive switches, which are interconnected
 in the most redundant way

0:15:34.320000 --> 0:15:41.300000
 possible. All right, so that's what you're
 going to find within your company.

0:15:41.300000 --> 0:15:45.220000
 You're going to find a combination of
 these last three, three tier, two

0:15:45.220000 --> 0:15:48.880000
 tier, and spine leaf
 architectures.

0:15:48.880000 --> 0:15:52.560000
 Now chances are though, the data at
 some point is going to have to leave

0:15:52.560000 --> 0:15:55.820000
 your company. You're going to want some
 data going out to the internet,

0:15:55.820000 --> 0:15:58.740000
 or maybe going to a partner's organization
 that's across the street or

0:15:58.740000 --> 0:16:01.860000
 across the city or in another
 different state or country.

0:16:01.860000 --> 0:16:04.640000
 So for that, we need WAN networks.


0:16:04.640000 --> 0:16:08.440000
 And just like those different architectures
 that we can build within our

0:16:08.440000 --> 0:16:12.400000
 company, like two tier, three tier,
 or spine leaf, there are different

0:16:12.400000 --> 0:16:14.560000
 WAN architectures as well.

0:16:14.560000 --> 0:16:19.320000
 So there's a variety of connection
 methods, and these basically break

0:16:19.320000 --> 0:16:23.600000
 down into three different
 architectures.

0:16:23.600000 --> 0:16:28.260000
 One architecture is point to point, which
 means that when you select this

0:16:28.260000 --> 0:16:34.460000
 architecture, that means you've got, for
 example, an office maybe in Boston.

0:16:34.460000 --> 0:16:38.300000
 And maybe another office
 in New York City.

0:16:38.300000 --> 0:16:43.980000
 If you want a point to point WAN connection
 between those two, then that

0:16:43.980000 --> 0:16:47.920000
 means that your telephone company or
 whoever sets up the point to point

0:16:47.920000 --> 0:16:51.080000
 connection is going to set it up in
 such a way that when you put data

0:16:51.080000 --> 0:16:56.920000
 on this wire right here, there's only
 one place it could possibly go over

0:16:56.920000 --> 0:16:58.120000
 to New York City.

0:16:58.120000 --> 0:17:01.640000
 And when he puts data on his wire,
 there's only one place it could go

0:17:01.640000 --> 0:17:06.900000
 over to Boston. There's only two
 points at the end of this WAN.

0:17:06.900000 --> 0:17:08.060000
 Boston, New York City.

0:17:08.060000 --> 0:17:10.280000
 That's why they call
 it point to point.

0:17:10.280000 --> 0:17:13.940000
 You say, well, Keith, what other options
 are there for me for WANs?

0:17:13.940000 --> 0:17:19.960000
 Well, you see, the downside to this
 architecture is if I call it.

0:17:19.960000 --> 0:17:24.320000
 So once again, let's say I had Boston
 here, and I had New York City here.

0:17:24.320000 --> 0:17:28.000000
 And if I call up my WAN provider and
 say, hey, I want another point to

0:17:28.000000 --> 0:17:34.720000
 point connection because I got another
 office coming online in Raleigh,

0:17:34.720000 --> 0:17:39.780000
 North Carolina. And I want
 to connect that up.

0:17:39.780000 --> 0:17:43.180000
 Okay, well, point to point means that's
 going to have to connect up to,

0:17:43.180000 --> 0:17:47.440000
 for example, to here to Boston, which
 means on Boston, I'm going to have

0:17:47.440000 --> 0:17:54.360000
 to have two physical cables coming in,
 and then, that's going to require

0:17:54.360000 --> 0:17:56.280000
 another interface on that router.

0:17:56.280000 --> 0:17:59.100000
 My Boston router is going to have another
 available interface with another

0:17:59.100000 --> 0:18:00.580000
 cable sticking out of it.

0:18:00.580000 --> 0:18:03.460000
 And both these cables are going to terminate
 like in a wall jack or something

0:18:03.460000 --> 0:18:06.720000
 at my office. Point to point.

0:18:06.720000 --> 0:18:10.060000
 So more cables, more connections.

0:18:10.060000 --> 0:18:12.020000
 Well, what other alternatives
 do we have?

0:18:12.020000 --> 0:18:16.660000
 We also have what's called
 a broadcast-based WAN.

0:18:16.660000 --> 0:18:20.120000
 For example, Metro Ethernet
 is like this.

0:18:20.120000 --> 0:18:25.520000
 With Metro Ethernet, we have a situation
 where you connect to the WAN

0:18:25.520000 --> 0:18:28.460000
 with some sort of a cable.

0:18:28.460000 --> 0:18:31.720000
 Okay, it's most likely going
 to be a fiber optic cable.

0:18:31.720000 --> 0:18:35.280000
 So this is a type of cable that's made
 out of like glass or plastic that's

0:18:35.280000 --> 0:18:38.620000
 carrying light. So instead of carrying
 electrical energy to represent

0:18:38.620000 --> 0:18:43.580000
 your data, light, like laser light, is
 going across that cable to represent

0:18:43.580000 --> 0:18:46.180000
 your data. Now, here's
 the big difference.

0:18:46.180000 --> 0:18:51.600000
 In this one right here, once your data
 gets put onto the fiber ring, everybody

0:18:51.600000 --> 0:18:53.880000
 connected to that
 ring will see it.

0:18:53.880000 --> 0:19:00.020000
 So let's say, for example, that in this
 particular case, customer A was

0:19:00.020000 --> 0:19:02.380000
 trying to send data to the ISP.

0:19:02.380000 --> 0:19:05.040000
 Well, once customer A puts the data on
 there and starts going around this

0:19:05.040000 --> 0:19:14.320000
 ring, broadcast-based WANs, or even
 broadcast-based LANs, we'll go to

0:19:14.320000 --> 0:19:20.940000
 customer C. That data will also go to
 customer B as well as going to the

0:19:20.940000 --> 0:19:24.800000
 ISP. It'll even circle all around again,
 and it'll be the job of customer

0:19:24.800000 --> 0:19:28.520000
 A to strip that data off the ring,
 to take it off of there.

0:19:28.520000 --> 0:19:31.180000
 That's a broadcast-based WAN.

0:19:31.180000 --> 0:19:33.240000
 Now, you might look at that
 and say, wait a second.

0:19:33.240000 --> 0:19:35.200000
 Why would I ever want to use that?


0:19:35.200000 --> 0:19:39.960000
 I mean, if I'm customer A, I don't want
 customer B and C to see my data.

0:19:39.960000 --> 0:19:41.260000
 Well, here's the nice thing.

0:19:41.260000 --> 0:19:46.720000
 In this type of WAN, these switches that
 you see connected here, you don't

0:19:46.720000 --> 0:19:48.040000
 actually own them.

0:19:48.040000 --> 0:19:50.160000
 The ISP owns them.

0:19:50.160000 --> 0:19:53.620000
 So all these switches are owned by the
 ISP, so whoever the building manager

0:19:53.620000 --> 0:19:58.760000
 is of this building and this building
 and this building, they will go

0:19:58.760000 --> 0:20:01.580000
 to the ISP and they'll say, hey, we
 want to connect to your broadcast

0:20:01.580000 --> 0:20:05.160000
-based WAN, your Metro Ethernet
 ring, for example.

0:20:05.160000 --> 0:20:10.460000
 And so the ISP will walk in and put a
 switch in each one of these locations

0:20:10.460000 --> 0:20:15.400000
 that connects up to this WAN, and then
 you, as the customer, you will

0:20:15.400000 --> 0:20:18.460000
 actually be in charge of taking your
 own network equipment, like maybe

0:20:18.460000 --> 0:20:23.500000
 you've got your own switch right here,
 and you'll connect it to this switch.

0:20:23.500000 --> 0:20:25.260000
 Now, I might say, well,
 how does that help me?

0:20:25.260000 --> 0:20:26.940000
 Here's how it helps you.

0:20:26.940000 --> 0:20:32.380000
 When your data goes right here and
 goes onto the WAN, yes, this switch

0:20:32.380000 --> 0:20:34.520000
 will see it. So let's just put
 another switch right here.

0:20:34.520000 --> 0:20:36.740000
 So this is actually
 owned by customer C.

0:20:36.740000 --> 0:20:39.280000
 This switch right here
 is owned by customer B.

0:20:39.280000 --> 0:20:42.180000
 This switch right here
 is owned by customer A.

0:20:42.180000 --> 0:20:44.360000
 And here's what's the difference.

0:20:44.360000 --> 0:20:51.280000
 Even though this, I'll just say right
 here, ISP, even though the ISP switch

0:20:51.280000 --> 0:20:56.480000
 in this building sees the data, it will
 recognize, oh, this data is not

0:20:56.480000 --> 0:20:58.720000
 actually meant for customer C.

0:20:58.720000 --> 0:21:01.380000
 So that data will stop right here.


0:21:01.380000 --> 0:21:07.020000
 So whoever, the actual customer who
 owns customer C switch won't see it.

0:21:07.020000 --> 0:21:09.720000
 Only the ISP switch will see it.

0:21:09.720000 --> 0:21:13.240000
 And the same thing, when it gets right
 here, yes, this ISP switch will

0:21:13.240000 --> 0:21:17.020000
 see it. But it will say, hey, it's
 not meant for this building.

0:21:17.020000 --> 0:21:19.440000
 So he'll sort of block
 it right there.

0:21:19.440000 --> 0:21:21.680000
 So customer B switch won't see it.


0:21:21.680000 --> 0:21:25.400000
 Once it gets over here, this ISP switch
 will say, oh, yeah, that's for

0:21:25.400000 --> 0:21:28.360000
 us. He'll pass it along through.

0:21:28.360000 --> 0:21:32.460000
 But it's still called a broadcast based
 WAN because every device that's

0:21:32.460000 --> 0:21:36.120000
 physically connected to
 it is seeing everything.

0:21:36.120000 --> 0:21:39.880000
 It just so happens that in most cases,
 the device is connected to this

0:21:39.880000 --> 0:21:42.820000
 type of WAN are owned
 by a service provider.

0:21:42.820000 --> 0:21:46.160000
 They're just placed in different buildings
 and sort of leased out or rented

0:21:46.160000 --> 0:21:49.180000
 out to different customers.

0:21:49.180000 --> 0:21:53.920000
 And then we have another type of WAN
 called a non-broadcast multi-access

0:21:53.920000 --> 0:22:04.620000
 WAN. This is a type of WAN where when you
 put your data onto it, you actually

0:22:04.620000 --> 0:22:07.540000
 address your data
 in a certain way.

0:22:07.540000 --> 0:22:14.080000
 And as that data enters the WAN, it will
 be kept separate from other people's

0:22:14.080000 --> 0:22:16.700000
 data and then go across.

0:22:16.700000 --> 0:22:23.040000
 So this is kind of like point to point
 in that if I had something like

0:22:23.040000 --> 0:22:27.960000
 this in point to point, well, any data
 going across here was never going

0:22:27.960000 --> 0:22:30.080000
 to be seen right here.

0:22:30.080000 --> 0:22:31.920000
 This guy was never
 going to see that.

0:22:31.920000 --> 0:22:34.600000
 Well, here's the difference.

0:22:34.600000 --> 0:22:39.920000
 In point to point WANs, I had to
 select which interface I wanted.

0:22:39.920000 --> 0:22:43.080000
 Did I want to put the data on this
 interface, which could only go over

0:22:43.080000 --> 0:22:47.280000
 here, or do I want to put the data on
 this interface, which would go right

0:22:47.280000 --> 0:22:52.180000
 here? Depending on which interface I
 selected, that's where the data went.

0:22:52.180000 --> 0:22:59.020000
 In a non-broadcast multi-access WAN,
 I only have one physical connection

0:22:59.020000 --> 0:23:05.520000
 to put my data on, but that data is
 addressed in a certain way that when

0:23:05.520000 --> 0:23:11.580000
 it hits the WAN provider, and here's
 our WAN provider right here, when

0:23:11.580000 --> 0:23:15.980000
 it hits the WAN provider, they have
 some super secret sauce to keep it

0:23:15.980000 --> 0:23:20.280000
 separated, so it will only go
 out to where it needs to go.

0:23:20.280000 --> 0:23:25.780000
 My data will never be seen by any other
 customers because non-broadcast

0:23:25.780000 --> 0:23:30.200000
 multi-access will prevent it.

0:23:30.200000 --> 0:23:35.320000
 So it's called multi-access because,
 let's just look at this here for

0:23:35.320000 --> 0:23:42.220000
 a second. So if this is the WAN here,
 this is probably a better picture.

0:23:42.220000 --> 0:23:49.220000
 Let's say this is customer A, let's
 say A has site 1, A has site 2, and

0:23:49.220000 --> 0:23:55.940000
 A has site 3. And let's say connected
 the same WAN is a totally different

0:23:55.940000 --> 0:24:03.000000
 company. It's got B1
 and B2 connected.

0:24:03.000000 --> 0:24:11.640000
 Well, when A1 sends the data, let's
 say he says, okay, I want this data

0:24:11.640000 --> 0:24:14.340000
 to go over to my other site, A2.

0:24:14.340000 --> 0:24:15.860000
 That's where I want this to be.

0:24:15.860000 --> 0:24:24.440000
 So this data will have a certain address
 on it, like let's say 101, and

0:24:24.440000 --> 0:24:30.460000
 these boxes here represent WAN switches
 owned by the service provider.

0:24:30.460000 --> 0:24:36.760000
 Now, once it hits the WAN switch, that
 WAN switch will say, oh, 101, that

0:24:36.760000 --> 0:24:41.440000
 means I need to switch it out like
 this way and send it to A2.

0:24:41.440000 --> 0:24:46.880000
 So first of all, that data will never
 pop out here at A3 because that

0:24:46.880000 --> 0:24:48.660000
 address didn't send it there.

0:24:48.660000 --> 0:24:54.520000
 It'll certainly not pop out here to
 B2, and it won't pop out here to B1.

0:24:54.520000 --> 0:24:58.000000
 The addressing will make sure it goes
 exactly where it needs to go.

0:24:58.000000 --> 0:25:03.860000
 Now, this particular cable right here
 could be used by other customers

0:25:03.860000 --> 0:25:06.120000
 as well, but you don't have
 to worry about that.

0:25:06.120000 --> 0:25:10.480000
 Even though this cable is transporting
 data from multiple customers, the

0:25:10.480000 --> 0:25:16.240000
 addresses in the data itself will keep
 the data separate and make sure

0:25:16.240000 --> 0:25:19.660000
 it never pops out where
 it's not supposed to go.

0:25:19.660000 --> 0:25:25.400000
 So the reason we call this multi-access
 is because this WAN cloud here

0:25:25.400000 --> 0:25:30.860000
 that I've drawn, this thing right
 here, is a multi-access cloud.

0:25:30.860000 --> 0:25:34.560000
 Lots of different devices can connect
 to it and use it to transport their

0:25:34.560000 --> 0:25:40.120000
 data. A1 can use this one connection
 right here to send data to location

0:25:40.120000 --> 0:25:42.800000
 A2 or location A3.

0:25:42.800000 --> 0:25:46.880000
 It's not like point-to-point where A1
 would need to have two cables, one

0:25:46.880000 --> 0:25:49.820000
 cable going to A2, another
 cable going to A3.

0:25:49.820000 --> 0:25:50.840000
 It's not like that.

0:25:50.840000 --> 0:25:55.760000
 With multi-access, we've got one cable
 that we can use to reach multiple

0:25:55.760000 --> 0:25:57.980000
 remote locations.

0:25:57.980000 --> 0:26:04.500000
 But this is non-broadcast multi-access,
 meaning there is no way we could

0:26:04.500000 --> 0:26:09.880000
 put data on this cable, one piece of
 data, and have it go out to A2 and

0:26:09.880000 --> 0:26:11.840000
 A3 simultaneously.

0:26:11.840000 --> 0:26:15.560000
 Can't do that, because when the data
 puts on this cable, we have to give

0:26:15.560000 --> 0:26:17.440000
 it some sort of an address.

0:26:17.440000 --> 0:26:20.200000
 The WAN provider will tell
 you what that address is.

0:26:20.200000 --> 0:26:23.460000
 And based on what address you put in
 that cable, in that data, it's going

0:26:23.460000 --> 0:26:25.820000
 to go to one location or another.

0:26:25.820000 --> 0:26:28.800000
 Maybe right behind this, I put another
 piece of data on the exact same

0:26:28.800000 --> 0:26:32.420000
 cable, maybe the address
 for this one is 301.

0:26:32.420000 --> 0:26:36.180000
 And when the WAN provider sees that,
 he knows, oh, okay, that needs to

0:26:36.180000 --> 0:26:38.540000
 go to location A3.

0:26:38.540000 --> 0:26:42.780000
 So this is non-broadcast
 multi-access.

0:26:42.780000 --> 0:26:52.640000
 Another type of an architecture
 is a small office, home office.

0:26:52.640000 --> 0:26:54.360000
 That's what SoHo stands for.

0:26:54.360000 --> 0:26:56.980000
 Here you can see on the left,
 you've got a small office.

0:26:56.980000 --> 0:26:58.980000
 On the right, you've
 got a home office.

0:26:58.980000 --> 0:27:03.560000
 These are typically smaller networks,
 so we have less equipment demands,

0:27:03.560000 --> 0:27:07.320000
 maybe just a single router
 or a single switch.

0:27:07.320000 --> 0:27:10.540000
 And less need for authentication and
 security, because you're in your

0:27:10.540000 --> 0:27:13.940000
 house, you're in your small office,
 you designed all this.

0:27:13.940000 --> 0:27:16.900000
 Very few people or devices
 are connecting to this.

0:27:16.900000 --> 0:27:21.200000
 The downside of this is it's difficult
 to manage and enforce a policy

0:27:21.200000 --> 0:27:24.100000
 from headquarters.

0:27:24.100000 --> 0:27:27.860000
 In other words, if I'm at corporate
 headquarters, I'm the network admin,

0:27:27.860000 --> 0:27:32.660000
 and I know that there's like 100 people
 that are working from home.

0:27:32.660000 --> 0:27:37.360000
 So there's 100 home office networks out
 there connecting back to my corporate

0:27:37.360000 --> 0:27:43.000000
 office. And there's like three or four
 small offices, like maybe branch

0:27:43.000000 --> 0:27:45.020000
 offices that are connecting
 back to me.

0:27:45.020000 --> 0:27:48.240000
 It's hard for me to say, okay,
 here's going to be our policy.

0:27:48.240000 --> 0:27:52.240000
 Everybody has to use this kind of authentication,
 this type of security.

0:27:52.240000 --> 0:27:55.660000
 Well, I can say that, but it's hard for
 me to enforce that, because those

0:27:55.660000 --> 0:28:00.420000
 100 people that are working from home,
 what mechanism do I have to ensure

0:28:00.420000 --> 0:28:04.180000
 that they've set up their little tiny
 network of like maybe one router,

0:28:04.180000 --> 0:28:08.980000
 and that's it, with the policy that
 I've told them they should set up.

0:28:08.980000 --> 0:28:12.600000
 Kind of difficult to enforce that.


0:28:12.600000 --> 0:28:17.000000
 And the last two architectures I want
 to talk about are sort of supersets

0:28:17.000000 --> 0:28:22.100000
 of the previous architectures we just
 looked at, on-premises versus cloud

0:28:22.100000 --> 0:28:24.120000
-based architecture.

0:28:24.120000 --> 0:28:29.760000
 So an on-premise network, that just
 simply means that your network and

0:28:29.760000 --> 0:28:34.460000
 all the devices, all the resources
 are right there at your building.

0:28:34.460000 --> 0:28:38.180000
 It's on your premises, all the cables,
 all the switches, all the routers,

0:28:38.180000 --> 0:28:41.120000
 all the servers, all the hosts,
 they're right there.

0:28:41.120000 --> 0:28:43.440000
 So you can see everything, you can
 touch everything, you're in control

0:28:43.440000 --> 0:28:49.280000
 of everything. Cloud-based network is
 a typical type of architecture where

0:28:49.280000 --> 0:28:51.300000
 a company might say,
 you know what?

0:28:51.300000 --> 0:28:58.820000
 What we need is like 25 different servers,
 because we have 25 different

0:28:58.820000 --> 0:29:01.820000
 kinds of data that people are
 needing access in my company.

0:29:01.820000 --> 0:29:08.460000
 Well, I could buy the 25 servers myself,
 then I would be responsible for

0:29:08.460000 --> 0:29:13.220000
 finding a place, a physical place in
 my building to put those things,

0:29:13.220000 --> 0:29:16.140000
 making sure I've got enough air conditioning
 in that room so they don't

0:29:16.140000 --> 0:29:20.360000
 overheat, determining the correct cabling
 to connect them out, then I

0:29:20.360000 --> 0:29:22.580000
 got to put the software on there.

0:29:22.580000 --> 0:29:26.620000
 If they ever crash, I got to troubleshoot
 that and fix that, that would

0:29:26.620000 --> 0:29:29.860000
 be on-premises. That would
 all be my responsibility.

0:29:29.860000 --> 0:29:31.560000
 Or I could say, you know what?

0:29:31.560000 --> 0:29:36.820000
 There's a company out there called
 Amazon AWS, and they've got servers

0:29:36.820000 --> 0:29:41.400000
 I can use. As long as I can get to the
 internet, I can rent servers from

0:29:41.400000 --> 0:29:46.020000
 them. So why don't I take my 25 servers
 and put them in the cloud instead,

0:29:46.020000 --> 0:29:47.900000
 have them reachable
 via the internet?

0:29:47.900000 --> 0:29:51.580000
 Then I don't have to worry about the
 physical server, AWS, they figure

0:29:51.580000 --> 0:29:55.520000
 that out. Amazon, they're responsible
 for if that server goes down fixing

0:29:55.520000 --> 0:29:58.900000
 it, they're responsible for patching
 it, they're responsible for figuring

0:29:58.900000 --> 0:30:01.440000
 out how that server is going to be
 able to get back to the internet so

0:30:01.440000 --> 0:30:02.720000
 I can get to it.

0:30:02.720000 --> 0:30:05.300000
 That's a cloud-based network.

0:30:05.300000 --> 0:30:10.200000
 So in a cloud-based network, the benefits
 of that are I could have multiple

0:30:10.200000 --> 0:30:14.420000
 offices, each office doesn't have to have
 their own set of servers, there's

0:30:14.420000 --> 0:30:18.360000
 just one common set of servers in the
 internet hosted by a company like

0:30:18.360000 --> 0:30:22.800000
 Amazon, and all my companies can get
 to those servers as long as they

0:30:22.800000 --> 0:30:24.060000
 have internet connectivity.

0:30:24.060000 --> 0:30:27.940000
 And it removes a lot of the responsibility
 from my shoulders and puts

0:30:27.940000 --> 0:30:31.680000
 it on the shoulders of the
 cloud service provider.

0:30:31.680000 --> 0:30:37.020000
 So that concludes this video on
 different network architectures.

0:30:37.020000 --> 0:30:37.940000
 I hope you found it useful.

