Host Scalability Part 3

00:04

Hello. Welcome to part three of the host scalability lesson in module 12. So now we're looking a little bit more At what d arrests conducive for us if we want to view the recommendations that dlrs gives if you're calling the last portion of the lesson we I looked at what the DRS information was

00:24

so I can select my my cluster in the inventory and select a DDOS tab.

00:30

Uh, you'll see a button in the upper right corner that lets you run, DRS.

00:35

And this will refresh your recommendation. So if you've made some changes and the DRS interval hasn't passed yet,

00:42

well, that might run every 15 minutes, for instance, every five minutes to paying and I've gotta configured you can refresh it manually.

00:49

There's also a link to let you out at the cluster properties if you need to modify the rules or into the other parameters that define your cluster

00:58

and because each of the recommendations will be in a separate line in the display, you can simply select

01:03

one or more of those

01:07

recommendations to get listed and run. Run those separately so you can run a subset of the recommendations

01:14

on the lower right side of the interface. You'll notice when we look at the lab. There's also a apply all recommendations button.

01:21

That way, if you want to, you look at the recommendations and I'll look reasonable. Just click, apply all and sit back and wait for it to complete.

01:30

The cluster status can also be modified or not modified but monitored. You go to select your cluster in the inventory and select the tasks and events button.

01:41

Once you do that, you'll get your list of tasks and events. There is a task or events button that you have to choose, though, so you look a tasks in one view and events in another view.

01:51

And this is really useful when you're troubleshooting or if you want to just see what happened with the recommendations in the past.

01:57

For instance, if you come into work one day, you see that V EMS are not in their usual places. You might go to your tasks and events

02:07

and look at the events button and see what actually happened. Why did this b m get moved?

02:10

It might be that VM one was moved to host number six

02:15

due to CPU and memory constraints, so you'll get that kind of detail there, and we'll see that again. Well, when we look at the lab

02:23

now, I need to talk about maintenance mode in standby mode.

02:28

You can right click on one of your hosts and going to maintenance mode.

02:31

You typically would do this when you're doing some maintenance on the host, maybe adding more memory if it's Ah um

02:38

ah, virtual host. Or, if you want to remove the host from the cluster, has to be a maintenance mode first.

02:46

So if you want to keep the VM is running, you have to migrate them somewhere else. Otherwise, they will get shut off and powered off when you put the host in maintenance mode.

02:55

Once the host is entering maintenance mode, you can't power on any V, ems or migrate that once you begin that process, you have to wait for a maintenance mode to complete and then make those decisions. So it's better to just decide ahead of time what to do with the PM's that are on a host that you want to put into maintenance mode.

03:15

Standby mode is effectively powering off the host, so if it's a physical host, and you want to add a network card at memory or ah or reboot it. Perhaps you can. You can do that. Reboot is also an option when you right click a host,

03:29

but that's not really a mode so much that's just more of an action.

03:34

So what's involved with removing a host of a cluster? We know that when we add a host to Acosta, we can just drag and drop.

03:39

It's very simple. You get a wizard that pops up asking about your resource pool and asking you

03:46

a couple other questions. We won't remove the host first reported in maintenance mode.

03:53

And if you remember, the resource pool hierarchy that was preserved by

03:57

when we move the host into the cluster will now be preserved. We pull the host back out,

04:01

so that will be visible in the inventory. Once the host gets pulled back out, all the PM's must be powered off because you're going into maintenance mode, same same as we discussed up here

04:14

and as a as a result of pulling the host out of the cluster. The available resource is for that cluster will decrease

04:20

if I have, you know four hosts. I pulled one out. Now I've only got three hosts to aggregate my memory and CP, your resource is

04:30

so some things to think about for how we improve the performance of virtual machines

04:34

starting at the fine under the scale we can. We can do things like do traffic shaping for the network

04:42

and this allows. You do specify things like maximum burst size and maximum bandwith that a network card can be allocated, and I'll explain that little bit more. We do the lab.

04:56

You can also change the CPU and memory reservations for that V M.

05:00

And this lets you find Tune. How many resource is a particular V Emily used relative to the other V ems that are running on that host.

05:09

If you have resource pools,

05:11

you could modify the CPU and memory limits and reservations for those resource pools.

05:16

So if you remember, we might have a production resource pool that gets 2/3 of our resource. Is that maybe testing development that gets 1/3 so you might have to adjust those sightings a little bit in order to get better performance.

05:30

So now we're getting monitors. The broad end of the scale. I can use Nick teaming. Remember, Nick Teaming only works with physical network cards, not with virtual network interfaces.

05:41

So if you have multiple network cards, you can team them

05:44

to get the balance, the load bouncing features of using two cards and having about redundancy as well.

05:51

We can have redundancy and load balancing with our storage by configuring multi path thing.

05:58

So I've got multiple past to get to the storage, and I could endure some failures in one of those paths, gets damaged or stops working.

06:04

And then the best option from the broad sense is to just create a dlrs cluster like we've been discussing.

06:11

This gives you the most flexibility for automatic load balancing automatic placement of E ems and lets you use some of your other features like resource pools and so on.

06:23

But let's the cluster aggregate resource is be used most efficiently.

06:29

So to recap,

06:30

we talked about what the recommendations mean, how you can look at them. How you can apply them is a subset, or apply the medicine an entire group.

06:39

You know that you can look at the status of the cluster by going to the tasks and events and leader clicking the task or event button.

06:46

We know what maintenance mode and standby motives for.

06:49

Remember, you have to power off the V EMS or move them somewhere if you want to use either one of those modes

06:56

pulling and close out. Pulling the host of a cluster involves putting in maintenance bill, which means you have to deal with the PM's power them off or move them.

07:04

And then we looked at some considerations for improving your performance of your V EMS.

07:12

All right, so you're next. Next task is to do lab number 21

07:15

in this lab. You'll be creating it in balance in your load

07:18

for your deal less cluster

07:20

uh,

07:21

you'll create the cluster.

07:24

You'll verify that DUS is working correctly,

07:28

and then you'll create affinity rules and and test those. You'll be able Thio create an anti finicky rule intestine

07:36

and then will create a host

07:40

to PM's rule.