WEBVTT
Kind: captions
Language: en

1
00:00:00.000 --> 00:00:06.149
Welcome to lesson eleven. Objective 1.6 asks you to compare and contrast network topologies,

2
00:00:06.149 --> 00:00:12.097
architectures and types. We'll cover three things. Topologies are the shape of the links:

3
00:00:12.097 --> 00:00:17.577
point to point, star, mesh and hybrid. Architectures are how switches are layered:

4
00:00:17.577 --> 00:00:23.859
the three-tier model, the collapsed core, and spine and leaf. And traffic flows describe which

5
00:00:23.859 --> 00:00:26.800
way traffic moves: north-south or east-west.

6
00:00:27.200 --> 00:00:34.115
Let's follow an online store. Its office and its warehouse are joined by a single dedicated link.

7
00:00:34.115 --> 00:00:40.958
That's a point-to-point topology: exactly two endpoints, and nothing else shares the connection.

8
00:00:40.958 --> 00:00:47.659
Leased lines, a direct link between two sites, and a wireless bridge between two buildings are

9
00:00:47.659 --> 00:00:54.218
all examples. Point to point is simple and private, but with one link there's only one path,

10
00:00:54.218 --> 00:00:56.000
so there's no redundancy.

11
00:00:56.400 --> 00:01:02.355
Inside the office, every computer plugs into one switch. That's a star topology:

12
00:01:02.355 --> 00:01:08.533
every device links to a central point. It's easy to add devices and easy to manage,

13
00:01:08.533 --> 00:01:15.307
but the center is a single point of failure. The wide area version is called hub and spoke:

14
00:01:15.307 --> 00:01:22.378
each branch connects only to the headquarters hub. Branch-to-branch traffic travels through the

15
00:01:22.378 --> 00:01:26.100
hub, which keeps routing simple but loads the hub.

16
00:01:26.500 --> 00:01:33.384
In a full mesh, every device has a direct link to every other device. If one link fails,

17
00:01:33.384 --> 00:01:40.816
traffic simply takes another path, so mesh gives the most redundancy. The cost is the number of

18
00:01:40.816 --> 00:01:46.448
links: n times n minus one, divided by two. Five routers need ten links,

19
00:01:46.448 --> 00:01:52.941
and ten routers would need forty-five. That's why many networks use a partial mesh,

20
00:01:52.941 --> 00:01:57.400
adding extra links only between the most important sites.

21
00:01:57.800 --> 00:02:05.575
Real networks rarely use just one shape. As the store grows, each site has its own star of computers

22
00:02:05.575 --> 00:02:11.173
around a switch, and the three sites are meshed together for redundancy.

23
00:02:11.173 --> 00:02:16.693
Combining two or more topologies like this is called a hybrid topology.

24
00:02:16.693 --> 00:02:22.213
On the exam, if a design mixes star and mesh, or any other combination,

25
00:02:22.213 --> 00:02:27.500
the answer is hybrid. Most real networks you'll work on are hybrids.

26
00:02:27.900 --> 00:02:33.255
Now architectures. A large campus often uses the three-tier hierarchical model.

27
00:02:33.255 --> 00:02:39.694
The access layer is where users and devices plug in, with features such as VLANs and power over

28
00:02:39.694 --> 00:02:46.133
Ethernet. The distribution layer aggregates the access switches and applies routing and policy,

29
00:02:46.133 --> 00:02:51.216
forming the boundary between Layer 2 and Layer 3. The core layer is a fast,

30
00:02:51.216 --> 00:02:56.300
highly available backbone that connects the distribution switches together.

31
00:02:56.700 --> 00:03:01.954
A smaller site doesn't need a separate core. In a collapsed core design,

32
00:03:01.954 --> 00:03:09.250
the core and distribution layers are merged into one pair of switches, and the access layer connects

33
00:03:09.250 --> 00:03:15.452
to them. That's a two-tier design: fewer switches, lower cost, and a simpler network,

34
00:03:15.452 --> 00:03:22.603
which suits a single building or a small campus. When a site grows to several buildings with their

35
00:03:22.603 --> 00:03:27.200
own distribution switches, adding a dedicated core makes sense.

36
00:03:27.600 --> 00:03:33.280
Data centers often use spine and leaf. Servers connect to leaf switches,

37
00:03:33.280 --> 00:03:39.196
and every leaf connects to every spine. Leaves don't connect to each other,

38
00:03:39.196 --> 00:03:46.532
and spines don't connect to each other. So a server on one leaf reaches a server on any other

39
00:03:46.532 --> 00:03:54.105
leaf through exactly one spine: leaf, spine, leaf. Every path is the same length and all of them

40
00:03:54.105 --> 00:04:00.100
can carry traffic. Need more ports? Add a leaf. More bandwidth? Add a spine.

41
00:04:00.500 --> 00:04:07.111
Engineers also describe which way traffic flows. North-south traffic enters or leaves the data

42
00:04:07.111 --> 00:04:12.104
center, such as a customer on the internet loading the store's website.

43
00:04:12.104 --> 00:04:19.207
East-west traffic stays inside, moving from server to server, like the web server asking the database

44
00:04:19.207 --> 00:04:25.888
for an order. Modern applications create a lot of east-west traffic, and spine and leaf handles

45
00:04:25.888 --> 00:04:30.600
it well because every server is the same distance from every other.

46
00:04:31.000 --> 00:04:37.724
On the exam, match the clue to the answer. Every device linked to every other is a full mesh.

47
00:04:37.724 --> 00:04:45.315
Everything connected to one central switch is a star, and branches that connect only through headquarters

48
00:04:45.315 --> 00:04:50.520
is hub and spoke. Core, distribution and access is the three-tier model,

49
00:04:50.520 --> 00:04:57.244
and merging core and distribution is a collapsed core. Every leaf connected to every spine is

50
00:04:57.244 --> 00:05:03.100
spine and leaf, and server-to-server traffic inside the data center is east-west.

51
00:05:03.500 --> 00:05:10.405
To recap: point to point joins exactly two endpoints. A star connects everything to a center,

52
00:05:10.405 --> 00:05:17.904
and hub and spoke does the same across a WAN. A mesh gives many paths, and a hybrid mixes topologies.

53
00:05:17.904 --> 00:05:24.884
The three-tier model has core, distribution and access layers, and a collapsed core merges the

54
00:05:24.884 --> 00:05:30.676
top two. Spine and leaf serves data centers, where east-west traffic is heavy.

55
00:05:30.676 --> 00:05:33.200
Next, we'll start IPv4 addressing.
