WEBVTT
Kind: captions
Language: en

1
00:00:00.000 --> 00:00:05.663
Our store's team has a plan for moving to AWS. Now the question is where,

2
00:00:05.663 --> 00:00:11.249
physically, the store will run. This lesson covers task 3.2 in Domain 3,

3
00:00:11.249 --> 00:00:18.386
the AWS global infrastructure. We'll cover two ideas. First, Regions and Availability Zones,

4
00:00:18.386 --> 00:00:25.756
and how they give you high availability. Second, edge locations, plus when to use more than one

5
00:00:25.756 --> 00:00:30.100
Region and a few ways to bring AWS even closer to users.

6
00:00:30.500 --> 00:00:36.296
An AWS Region is a separate geographic area, such as Northern Virginia or Frankfurt.

7
00:00:36.296 --> 00:00:42.645
Each Region is isolated from the others, and most resources you create belong to one Region.

8
00:00:42.645 --> 00:00:49.062
Inside every Region are multiple Availability Zones, and AWS builds each Region with at least

9
00:00:49.062 --> 00:00:55.480
three. Each Availability Zone is one or more discrete data centers. So the nesting is simple:

10
00:00:55.480 --> 00:01:01.000
Regions contain Availability Zones, and Availability Zones contain data centers.

11
00:01:01.400 --> 00:01:07.934
What makes an Availability Zone special is independence. Each one has its own redundant power,

12
00:01:07.934 --> 00:01:13.079
networking and connectivity, in separate facilities many kilometers apart,

13
00:01:13.079 --> 00:01:20.030
so a fire or a flood affects only one zone. Availability Zones don't share single points of failure.

14
00:01:20.030 --> 00:01:26.078
Yet they're linked by fast, redundant fiber, and AWS encrypts all traffic between them.

15
00:01:26.078 --> 00:01:31.500
That link is quick enough to copy data synchronously from one zone to another.

16
00:01:31.900 --> 00:01:39.251
That independence is how you get high availability. Our store runs a web server in Availability

17
00:01:39.251 --> 00:01:46.524
Zone A and another in Availability Zone B. A load balancer in front spreads shoppers' requests

18
00:01:46.524 --> 00:01:53.566
across both. The database keeps a primary copy in one zone and a standby copy in the other,

19
00:01:53.566 --> 00:02:00.917
kept in sync. Nothing important lives in only one place, so no single zone failure can take the

20
00:02:00.917 --> 00:02:02.000
store offline.

21
00:02:02.400 --> 00:02:09.806
Now say Availability Zone B loses power. The load balancer's health checks notice that the server

22
00:02:09.806 --> 00:02:16.983
there isn't responding, so it sends every request to zone A instead. If the database's primary

23
00:02:16.983 --> 00:02:23.167
copy had been in zone B, the standby would take over. Shoppers keep checking out.

24
00:02:23.167 --> 00:02:28.893
This is a key exam idea: to survive the failure of a data center or a zone,

25
00:02:28.893 --> 00:02:32.100
deploy across multiple Availability Zones.

26
00:02:32.500 --> 00:02:37.874
Which Region should the store use? Four factors decide. First, compliance:

27
00:02:37.874 --> 00:02:43.248
if a law says data must stay in a country, that overrides everything else.

28
00:02:43.248 --> 00:02:49.130
Second, latency: pick a Region close to your users. Third, services and features:

29
00:02:49.130 --> 00:02:54.649
not every service is in every Region, and new ones arrive gradually. Fourth,

30
00:02:54.649 --> 00:03:00.967
cost: prices differ between Regions. Most of our store's shoppers are in North America,

31
00:03:00.967 --> 00:03:03.000
so it picks a nearby Region.

32
00:03:03.400 --> 00:03:10.057
Most workloads run well in one Region. The exam guide names four reasons to use more than one.

33
00:03:10.057 --> 00:03:16.147
Disaster recovery, with a copy ready in a second Region if a whole Region has trouble.

34
00:03:16.147 --> 00:03:21.741
Business continuity, keeping critical operations running through a major event.

35
00:03:21.741 --> 00:03:28.327
Low latency for users on another continent. And data sovereignty, keeping each country's data

36
00:03:28.327 --> 00:03:33.567
in that country. AWS doesn't copy your data between Regions automatically;

37
00:03:33.567 --> 00:03:34.700
you set that up.

38
00:03:35.100 --> 00:03:41.711
Now the edge. Edge locations are AWS sites in many cities, far more of them than Regions,

39
00:03:41.711 --> 00:03:48.843
placed close to users. You don't run your servers there. Amazon CloudFront caches copies of your

40
00:03:48.843 --> 00:03:54.934
content at edge locations, so a shopper in Europe gets product images from nearby,

41
00:03:54.934 --> 00:04:01.769
and only cache misses travel back to the Region. Route 53 answers DNS queries from the edge,

42
00:04:01.769 --> 00:04:08.900
Global Accelerator brings traffic onto the AWS network, and Shield and WAF filter attacks there.

43
00:04:09.300 --> 00:04:15.600
Three more options bring AWS even closer. Local Zones extend a Region into a large city,

44
00:04:15.600 --> 00:04:22.044
for applications that need single-digit millisecond latency, like gaming or video editing.

45
00:04:22.044 --> 00:04:29.275
Wavelength Zones put AWS compute inside telecom 5G networks, for ultra-low latency to mobile devices.

46
00:04:29.275 --> 00:04:34.645
And AWS Outposts puts AWS-managed racks or servers in your own data center,

47
00:04:34.645 --> 00:04:39.800
with the same APIs and tools. Each one is still tied to a parent Region.

48
00:04:40.200 --> 00:04:45.312
Here's how the exam asks about this. Keep running if a data center fails:

49
00:04:45.312 --> 00:04:51.053
multiple Availability Zones. Survive the loss of a whole Region: multiple Regions.

50
00:04:51.053 --> 00:04:56.305
Data that must stay in one country: choose the Region for data sovereignty.

51
00:04:56.305 --> 00:05:01.627
Cache content close to users around the world: CloudFront at edge locations.

52
00:05:01.627 --> 00:05:07.859
Single-digit millisecond latency in a big city: Local Zones. AWS in your own data center:

53
00:05:07.859 --> 00:05:11.500
Outposts. Ultra-low latency for 5G apps: Wavelength.

54
00:05:11.900 --> 00:05:18.363
To recap: a Region is a separate geographic area with multiple isolated Availability Zones,

55
00:05:18.363 --> 00:05:24.115
and each zone is one or more data centers with no shared single point of failure.

56
00:05:24.115 --> 00:05:30.152
Use multiple zones for high availability, and multiple Regions for disaster recovery,

57
00:05:30.152 --> 00:05:36.259
distant users and data sovereignty. Edge locations cache and route traffic near users.

58
00:05:36.259 --> 00:05:41.727
Next, we'll look at the ways to work with AWS: the console, the command line,

59
00:05:41.727 --> 00:05:44.000
SDKs and infrastructure as code.
