5.2 vSphere High Availability (HA)

00:00

Hi,

00:01

you're watching Model fire

00:03

piece for advanced features.

00:05

Listen to the sphere. High availability or H A.

00:10

By the end of this video, you'll understand h a operation and how to configure it.

00:16

These were high availability is a future that ants resilience to your data center.

00:21

In fact, it protects your virtual machines against four problems.

00:26

Host failure, guest operating system failure, application failure and storage failure.

00:33

A. J restarts virtual machines on another host when the Hearst they are running on fails.

00:40

Pretty cool in this video, and for the sake of simplicity, we are going to discuss how h A protects your V EMS against host failures only

00:50

first of all, in order to use H. A. You need a cluster of at least two years Excite House and no more than 64.

00:58

You also need a static management I p address on all your hosts. At least one management network, although it is highly recommended to have to

01:07

the virtual machines, you want to be protected, need to be on the shared storage,

01:12

and you need to configure NTP on all hosts.

01:17

Clock synchronization is actually required. No

01:21

A J is enabled on a cluster object in V center, so it actually requires the creation of a cluster. Now, once you satisfy all these requirements, your yes, Excite hosts will elect a master host.

01:34

How and why they do this is outside the scope of this lesson.

01:38

But the master host is the only host responsible for protecting your virtual machines.

01:44

Now, how does that work?

01:46

Well, the master host expects heartbeats coming from all the hosts in the cluster and vice versa. All the host expect the heartbeat of the master as well. Now these heartbeats are an indication that the host is alive and well.

02:01

When I host fails, it can no longer send heartbeats.

02:06

The master host detects that, and then it considers the failed host as dead. The master host then restarts the PM's that were running on the failed host on another host.

02:16

Then there is a dilemma here.

02:19

What if the host is still alive?

02:22

But it can't send heartbeats because of a problem in its management? VM Nick.

02:27

Now, first of all, that's why it is recommended to have two heartbeat networks

02:31

now, in the unlikely event that both heartbeat networks fam a cherry lets you configure a data store for heartbeats,

02:39

this state of storms to be a shared data store between all your hosts and to which they will heartbeat constantly.

02:46

So the host that loses its management, BMX, can still heartbeat to the data store using the storage network.

02:54

And that's how the master host knows if a host is really dead or it's just a networking problem.

03:01

Now there's a lot more trophies for A J, and we can spend a whole chapter talking about it. But in this lesson, I just want you to understand what a J does.

03:12

Let's now jump to the center and configure the thing. The first thing you want to do is create a cluster object. Remember, Peace? Fear A. J is enabled on a cluster off hosts. To do so right, click the data center object and choose new cluster.

03:27

Give it a name. I'm gonna name it Computer cluster. Then check the box that says turn all next to these here H A.

03:36

Now, if you unhygienic the menu under these very J, you can enable or disable specific options.

03:43

Just leave it like it is and click Okay.

03:45

The next step is to add our two hosts to the newly created cluster.

03:50

You do that by just dragging and dropping them inside the cluster.

03:54

The hosts will display a warning telling you that your hosts have no redundancy in the management network.

04:01

This is what we talked about earlier. It is highly recommended to have to management networks for heartbeat redundancy.

04:09

In our case, this is just a proof of concept. So we will leave it like this.

04:15

No.

04:15

If you right, click the cluster and choose settings

04:18

and these fear availability. Then click on edit.

04:23

You lend in the main wizard where you configure everything related to these very H A

04:29

now covering every feature and what it does is outside the scope off the course. What we will do instead is configured Ah, Heartbeat data store. And you do that by navigating to the heartbeat Data stores tab

04:44

and in the bottom pain you'll see all the shared data stored in your data center.

04:49

It is recommended to choose to data stores were heartbeat. So let's select the only two data stores we've got.

04:56

Let's actually test the sphere J. We're going to simulate the failure off. Yes, excite too, and see what happens. But before we do that, let me remind you that a will not be able to protect the EMS that are not only shared storage. In our case, yes, Excite, too, is hosting the two database PM's.

05:15

Let's make sure that both are being stored in the share to data store.

05:19

Click on the database VM

05:23

and then on the data stores. Tap in the middle pain

05:26

and it is only shared storage.

05:29

Okay, lets very fine database, too,

05:31

and it's on the local storage. So let's migrated to a shared story it using story tree motion.

05:39

So right click of'em and choose my great

05:43

select Change storage. Only

05:46

Then choose the ice Cosy State a store

05:50

and it's now one shared data store.

05:53

Let's now simulate host failure by shutting down his ex. I, too,

05:58

You do that by right, clicking the host

06:00

and go to power and then shut down.

06:03

It will prompt you to enter a reason for shutting down the host. So just type testing this year. A chai and click. Okay,

06:12

it could take a while.

06:15

Well again, speed up the video until the host becomes unresponsive.

06:19

There we are.

06:23

You can see here that although yes excited, too, is powered off, the two database VM are powered on.

06:30

That's because of this fear A re started them on Essex. I won upon the failure off years Excited, too.

06:38

And if I click on the database PM, for example, you can see here that it is running on the first host

06:44

and the same goes for database, too.

06:47

Impressive, isn't it?

06:49

That was a quick demonstration of Easter H A.