WEBVTT
Kind: captions
Language: en

1
00:00:00.000 --> 00:00:06.963
Welcome to lesson five, where we finish objective 1.3. Last time, our online store built a virtual

2
00:00:06.963 --> 00:00:14.351
private cloud. Now the office needs a private way to reach it. First, we'll compare the two connectivity

3
00:00:14.351 --> 00:00:21.314
options on the exam: a site-to-site VPN and Direct Connect. Then we'll step back and look at cloud

4
00:00:21.314 --> 00:00:27.211
deployment models and service models, and three traits the exam loves: scalability,

5
00:00:27.211 --> 00:00:29.200
elasticity and multitenancy.

6
00:00:29.600 --> 00:00:36.447
The quickest option is a site-to-site VPN. The office router, called the customer gateway device,

7
00:00:36.447 --> 00:00:43.153
builds an encrypted IPsec tunnel across the public internet to a virtual private gateway on the

8
00:00:43.153 --> 00:00:48.165
cloud side. In AWS, each VPN connection has two tunnels for redundancy.

9
00:00:48.165 --> 00:00:54.659
It's cheap and can be running within hours. The catch: the traffic still rides the internet,

10
00:00:54.659 --> 00:00:58.400
so speed and latency change with internet conditions.

11
00:00:58.800 --> 00:01:05.611
Direct Connect is the other option. Instead of the internet, a dedicated fiber connection runs

12
00:01:05.611 --> 00:01:12.205
from the office network to a Direct Connect location, where it meets the provider's router.

13
00:01:12.205 --> 00:01:18.292
Traffic bypasses internet providers, so latency is consistent and bandwidth is high,

14
00:01:18.292 --> 00:01:25.103
with dedicated ports of one to four hundred gigabits per second. Setting it up takes weeks and

15
00:01:25.103 --> 00:01:30.900
costs more. And it isn't encrypted by default: add MACsec, or run a VPN over it.

16
00:01:31.300 --> 00:01:37.276
Side by side: a VPN travels over the public internet, with encryption built in,

17
00:01:37.276 --> 00:01:44.009
quick to set up and low in cost, but its performance varies. Direct Connect is a private,

18
00:01:44.009 --> 00:01:49.985
dedicated link with consistent performance, but it takes longer and costs more,

19
00:01:49.985 --> 00:01:56.869
and you add encryption yourself. Many companies use both: Direct Connect for daily traffic,

20
00:01:56.869 --> 00:02:00.500
and a VPN as the backup link if the fiber fails.

21
00:02:00.900 --> 00:02:08.280
Now, where does the cloud run? NIST defines the models. A public cloud is open to the general

22
00:02:08.280 --> 00:02:14.072
public and runs in the provider's data centers, shared by many customers.

23
00:02:14.072 --> 00:02:21.531
A private cloud is for one organization only, on or off its premises, giving more control at a

24
00:02:21.531 --> 00:02:29.228
higher cost. A hybrid cloud combines two or more clouds, linked so data and apps can move between

25
00:02:29.228 --> 00:02:35.100
them. Our office plus its VPC, joined by VPN or Direct Connect, is hybrid.

26
00:02:35.500 --> 00:02:42.213
Service models answer a different question: who manages what? With infrastructure as a service,

27
00:02:42.213 --> 00:02:48.926
or IaaS, the provider runs the hardware and virtualization, and you manage the operating system

28
00:02:48.926 --> 00:02:55.639
and everything above it. With platform as a service, or PaaS, you just deploy your application;

29
00:02:55.639 --> 00:03:01.222
the provider runs the operating system and runtime. With software as a service,

30
00:03:01.222 --> 00:03:06.804
or SaaS, you simply use a finished application, like web email. In every model,

31
00:03:06.804 --> 00:03:08.500
you still own your data.

32
00:03:08.900 --> 00:03:14.316
Two traits sound alike. Scalability is the ability to grow to meet demand.

33
00:03:14.316 --> 00:03:19.585
You can scale up, or vertically, by giving a server more CPU and memory,

34
00:03:19.585 --> 00:03:25.147
or scale out, horizontally, by adding more servers. Elasticity goes further:

35
00:03:25.147 --> 00:03:31.148
capacity grows and shrinks automatically as demand changes. During a holiday sale,

36
00:03:31.148 --> 00:03:38.100
the store adds servers, then removes them when traffic drops, so it pays only for what it uses.

37
00:03:38.500 --> 00:03:43.367
Finally, multitenancy. In a public cloud, many customers, called tenants,

38
00:03:43.367 --> 00:03:49.500
share the same physical servers. Our store's virtual machine may sit next to a competitor's.

39
00:03:49.500 --> 00:03:55.767
The hypervisor and the provider's controls keep each tenant isolated. NIST calls this resource

40
00:03:55.767 --> 00:04:02.433
pooling, and it's why the cloud is affordable. The risks are noisy neighbors using shared resources,

41
00:04:02.433 --> 00:04:06.100
and any flaw that breaks the isolation between tenants.

42
00:04:06.500 --> 00:04:13.458
On the exam, match the clue to the answer. An encrypted tunnel over the internet is a site-to-site

43
00:04:13.458 --> 00:04:18.570
VPN. A dedicated private link with consistent latency is Direct Connect.

44
00:04:18.570 --> 00:04:24.960
On-premises systems joined with a public cloud is hybrid. You manage the operating system:

45
00:04:24.960 --> 00:04:30.214
IaaS. You deploy code onto a managed platform: PaaS. You just use the app:

46
00:04:30.214 --> 00:04:34.900
SaaS. And adding and removing servers automatically is elasticity.

47
00:04:35.300 --> 00:04:41.231
To recap: a site-to-site VPN is an encrypted tunnel over the internet, quick and cheap.

48
00:04:41.231 --> 00:04:46.139
Direct Connect is a private, dedicated link with consistent performance.

49
00:04:46.139 --> 00:04:51.389
Clouds are public, private or hybrid. With IaaS you run the operating system,

50
00:04:51.389 --> 00:04:57.047
with PaaS you run the app, and with SaaS you just use it. Scalability means growth,

51
00:04:57.047 --> 00:05:03.114
elasticity means automatic growing and shrinking, and multitenancy means shared hardware.

52
00:05:03.114 --> 00:05:07.000
Next, we start objective 1.4: common ports and protocols.
