1
00:00:01,140 --> 00:00:03,440
Cluster Load Balancing.

2
00:00:03,440 --> 00:00:04,240
You might think, Tim,

3
00:00:04,240 --> 00:00:07,820
why are we looking at load balancing from a different perspective?

4
00:00:07,820 --> 00:00:10,740
Isn't this what failover clustering is all about?

5
00:00:10,740 --> 00:00:12,960
Well, kind of.

6
00:00:12,960 --> 00:00:17,030
I mean, it depends is the classic IT consultant's answer.

7
00:00:17,030 --> 00:00:21,560
Recall that failover clustering itself is a way to do high availability,

8
00:00:21,560 --> 00:00:23,050
and yes, as we saw,

9
00:00:23,050 --> 00:00:27,340
there are some features now and then that crop up that help with load balancing,

10
00:00:27,340 --> 00:00:33,040
that automatic client rebalancing feature that's part of SOFS as an example,

11
00:00:33,040 --> 00:00:36,850
but the exam wants you to understand some other features.

12
00:00:36,850 --> 00:00:38,340
For instance,

13
00:00:38,340 --> 00:00:42,650
there is a property called auto balancer level that you can set on your

14
00:00:42,650 --> 00:00:46,540
cluster that optimizes cluster node utilizations.

15
00:00:46,540 --> 00:00:47,440
Again,

16
00:00:47,440 --> 00:00:52,400
this is where clustered VMs may be moved due to patching

17
00:00:52,400 --> 00:00:56,560
and also to support that idea of preventing an unbalanced

18
00:00:56,560 --> 00:00:58,950
cluster where one cluster node,

19
00:00:58,950 --> 00:01:02,600
and this is different from the SOFS automatic client rebalancing,

20
00:01:02,600 --> 00:01:07,300
by the way, this failover clustering VM load balancing totally separate property,

21
00:01:07,300 --> 00:01:10,240
but it has some overlap in functionality.

22
00:01:10,240 --> 00:01:14,740
We're looking to make sure that we can support our cluster nodes so that they

23
00:01:14,740 --> 00:01:19,040
don't wind up getting oversaturated hosting too many VMs.

24
00:01:19,040 --> 00:01:24,040
How does the failover cluster implement this built‑in VM load balancing feature?

25
00:01:24,040 --> 00:01:27,040
Well there are some heuristics that play, some metrics.

26
00:01:27,040 --> 00:01:31,270
Basically, the cluster is looking at the host's memory pressure,

27
00:01:31,270 --> 00:01:33,740
as well as their CPU utilization.

28
00:01:33,740 --> 00:01:36,100
And then from a configuration standpoint,

29
00:01:36,100 --> 00:01:40,140
you can use PowerShell to set the AutoBalancerLevel property.

30
00:01:40,140 --> 00:01:44,120
The default setting is 1 in which the cluster will live

31
00:01:44,120 --> 00:01:47,580
migrate a VM to a different host when the host is greater

32
00:01:47,580 --> 00:01:50,500
than 80% utilized with CPU and RAM.

33
00:01:50,500 --> 00:01:52,330
You can also switch that.

34
00:01:52,330 --> 00:01:55,420
There are a couple other legal values here where you can move

35
00:01:55,420 --> 00:01:58,940
VMs when the host is 70% or more utilized,

36
00:01:58,940 --> 00:02:03,780
and then there is an option 3 that'll live migrate your VMs when the host is

37
00:02:03,780 --> 00:02:08,840
just greater than 5% above the aggregate cluster average.

38
00:02:08,840 --> 00:02:10,370
So you would set this,

39
00:02:10,370 --> 00:02:14,940
you'd either leave the property at the default or change it depending upon how

40
00:02:14,940 --> 00:02:19,030
aggressive you want the cluster to be in protecting nodes.

41
00:02:19,030 --> 00:02:21,510
So that is a property you should know for your exam,

42
00:02:21,510 --> 00:02:28,000
very useful when you're doing highly available virtual machines in your failover cluster.