1
00:00:00,840 --> 00:00:05,630
This next slide has one of my favorite Microsoft Docs diagrams,

2
00:00:05,630 --> 00:00:10,400
and it shows an implementation of ASR in a, well, we could look

3
00:00:10,400 --> 00:00:14,580
at it, it's mentioned in the small text here, Primary Site

4
00:00:14,580 --> 00:00:16,370
(On‑Premise). That's incorrect.

5
00:00:16,370 --> 00:00:20,730
It should be On‑Premises. But we just as well can look at this topology as

6
00:00:20,730 --> 00:00:25,680
Azure to Azure. Let's pretend that on the left our primary site is an Azure

7
00:00:25,680 --> 00:00:29,820
virtual network running in our primary region, and then our VMs that

8
00:00:29,820 --> 00:00:33,850
comprise our workload on the right, our failover site, may just as well be a

9
00:00:33,850 --> 00:00:36,440
VNet in another Azure region.

10
00:00:36,440 --> 00:00:39,750
I do want to say though, again though, you can go from on‑prem to Azure,

11
00:00:39,750 --> 00:00:43,870
from Azure to on‑prem, from Azure to Azure. It's very flexible. But the

12
00:00:43,870 --> 00:00:47,930
idea of the workload means that you're replicating not one, but all the

13
00:00:47,930 --> 00:00:52,120
machines that make up your workload, and you're replicating the VM

14
00:00:52,120 --> 00:00:56,860
configuration, which isn't that large, but principally the storage,

15
00:00:56,860 --> 00:01:00,340
which is large, and that's going to be the lion's share of the

16
00:01:00,340 --> 00:01:04,750
time, getting the storage on those machines synchronized from

17
00:01:04,750 --> 00:01:09,330
your primary site to your Azure failover site, right? So that is

18
00:01:09,330 --> 00:01:11,710
some of the mechanics. Now, you might be thinking,

19
00:01:11,710 --> 00:01:15,920
well, okay, the private IP addresses that we're using in the primary site,

20
00:01:15,920 --> 00:01:19,400
we're going to have to duplicate those in the failover site because

21
00:01:19,400 --> 00:01:21,840
we want to mirror the environment, right? And yes,

22
00:01:21,840 --> 00:01:22,600
that is right.

23
00:01:22,600 --> 00:01:26,870
Remember in Azure that you can absolutely have overlapping address ranges

24
00:01:26,870 --> 00:01:31,060
on VNets because VNets are air gapped from each other.

25
00:01:31,060 --> 00:01:33,420
Of course, you're not going to be able to create a peering

26
00:01:33,420 --> 00:01:36,540
between VNets that have overlapping addresses.

27
00:01:36,540 --> 00:01:38,550
But a big question would be, well,

28
00:01:38,550 --> 00:01:42,850
let's say that this is an Azure‑to‑Azure situation, and our primary site is

29
00:01:42,850 --> 00:01:47,590
in East US, and our failover site is in West Us, and there is a failure on

30
00:01:47,590 --> 00:01:50,090
Microsoft's side where the primary goes down.

31
00:01:50,090 --> 00:01:52,480
How in the world are we going to scramble?

32
00:01:52,480 --> 00:01:54,170
What exactly happens?

33
00:01:54,170 --> 00:01:54,370
Well,

34
00:01:54,370 --> 00:01:57,890
this solution is not required, but it's a really clever one

35
00:01:57,890 --> 00:02:00,120
where we're using Azure Traffic Manager.

36
00:02:00,120 --> 00:02:05,480
This is an OSI layer 7 DNS‑based load balancer, and we would use

37
00:02:05,480 --> 00:02:08,310
this as the ultimate endpoint for our application.

38
00:02:08,310 --> 00:02:09,240
Why?

39
00:02:09,240 --> 00:02:12,100
Well, we can place both of these deployments,

40
00:02:12,100 --> 00:02:16,480
that is our primary and our failover, behind the Traffic Manager's

41
00:02:16,480 --> 00:02:20,040
endpoints, and we can configure Traffic Manager to route all

42
00:02:20,040 --> 00:02:23,690
traffic to our primary and only use the secondary in the event

43
00:02:23,690 --> 00:02:26,540
that the primary is unreachable, you see.

44
00:02:26,540 --> 00:02:28,330
So that's a really nice solution.

45
00:02:28,330 --> 00:02:32,220
If, for instance, you have regional application gateways or load balancers

46
00:02:32,220 --> 00:02:39,000
in front of each of your environments, again, you would plug those in as endpoints in your Traffic Manager, and away you go