WEBVTT

0:00:03.020000 --> 0:00:07.520000
 In this video, we're going to take a
 look at virtual machine scale sets.

0:00:07.520000 --> 0:00:10.380000
 The topics that we're going to cover
 include what virtual machine scale

0:00:10.380000 --> 0:00:12.120000
 sets are, some basic concepts.

0:00:12.120000 --> 0:00:17.060000
 We're going to talk about auto scale, what
 auto scale is, why it's important.

0:00:17.060000 --> 0:00:19.220000
 And then also, we'll talk
 about load balancing.

0:00:19.220000 --> 0:00:21.320000
 All of these really go together.

0:00:21.320000 --> 0:00:25.220000
 And finally, I'm going to demonstrate
 a virtual machine scale set.

0:00:25.220000 --> 0:00:28.360000
 So let's go ahead and
 let's jump into this.

0:00:28.360000 --> 0:00:31.500000
 When I think about a virtual machine
 scale set, you need to think about

0:00:31.500000 --> 0:00:34.120000
 what the purpose of this is.

0:00:34.120000 --> 0:00:35.160000
 And here's the idea.

0:00:35.160000 --> 0:00:36.820000
 Let's say that you've
 got a workload.

0:00:36.820000 --> 0:00:42.620000
 And I've got a workload and I want
 some redundancy for my workload.

0:00:42.620000 --> 0:00:45.480000
 So I've got three virtual machines
 and let's say it's a web app.

0:00:45.480000 --> 0:00:52.160000
 They're all running
 the same web app.

0:00:52.160000 --> 0:01:00.200000
 And if I'm going to have any kind of
 actual availability, I want a load

0:01:00.200000 --> 0:01:03.200000
 balancer in front of that and also
 the load balancer itself has to be

0:01:03.200000 --> 0:01:03.960000
 highly available.

0:01:03.960000 --> 0:01:10.280000
 And you can do this exactly like this
 in Azure, no problem using virtual

0:01:10.280000 --> 0:01:13.160000
 machines, a load balancer,
 and that all works fine.

0:01:13.160000 --> 0:01:18.280000
 The problem is that it's a little
 bit difficult to reproduce.

0:01:18.280000 --> 0:01:24.080000
 So let's say that I want to
 go up to four instances.

0:01:24.080000 --> 0:01:30.380000
 Well, not only do I need the virtual
 machine, I actually need the NIC

0:01:30.380000 --> 0:01:34.460000
 associated with the virtual machine and
 I need to configure the load balancer.

0:01:34.460000 --> 0:01:38.520000
 Oh, and by the way, I also need to configure
 whatever's going to go in,

0:01:38.520000 --> 0:01:43.880000
 whatever settings are going to
 go into that new instance.

0:01:43.880000 --> 0:01:46.460000
 And so there's a lot
 of manual activity.

0:01:46.460000 --> 0:01:53.180000
 Also what happens when these are all
 maxed out, they're all at 100% across

0:01:53.180000 --> 0:01:58.480000
 the board. Then I have to go and manually
 say, oh no, all of my machines

0:01:58.480000 --> 0:02:01.640000
 are at 100% because that is
 exactly how I would say it.

0:02:01.640000 --> 0:02:03.900000
 And I would have to manually
 add another one.

0:02:03.900000 --> 0:02:09.940000
 Or likewise, what if we have all these
 machines and we're at 20% capacity.

0:02:09.940000 --> 0:02:14.300000
 I could look at that and I could say,
 oh no, we're wasting lots of money

0:02:14.300000 --> 0:02:16.860000
 because we have all these machines
 that aren't doing anything.

0:02:16.860000 --> 0:02:20.980000
 And I'd have to go in there and I would
 have to manually add or remove

0:02:20.980000 --> 0:02:27.420000
 instances. And so while that's absolutely
 possible, it's in no way optimal.

0:02:27.420000 --> 0:02:30.540000
 There's a lot of manual
 interaction.

0:02:30.540000 --> 0:02:31.780000
 This is not automated.

0:02:31.780000 --> 0:02:37.640000
 It can't really be responsive to changes
 in load, not certainly again

0:02:37.640000 --> 0:02:38.700000
 in an automated way.

0:02:38.700000 --> 0:02:42.880000
 So to get around all of this, we have
 what's called the virtual machine

0:02:42.880000 --> 0:02:47.280000
 scale set. And the virtual machine
 scale set simplifies the management

0:02:47.280000 --> 0:02:49.860000
 of a scalable workload.

0:02:49.860000 --> 0:02:55.740000
 With a virtual machine scale
 set, we have one resource.

0:02:55.740000 --> 0:02:58.980000
 That is my virtual
 machine scale set.

0:02:58.980000 --> 0:03:06.820000
 Now, associated with the virtual machine
 scale set are a set of instances.

0:03:06.820000 --> 0:03:11.740000
 And these instances are in
 fact virtual machines.

0:03:11.740000 --> 0:03:15.940000
 But they're managed, I'm going to say
 I for instance, they're managed

0:03:15.940000 --> 0:03:19.220000
 through a single virtual
 machine scale set.

0:03:19.220000 --> 0:03:25.680000
 And the virtual machine scale set is associated,
 you can draw that properly,

0:03:25.680000 --> 0:03:29.420000
 with a load balancer.

0:03:29.420000 --> 0:03:31.520000
 And it's automatic.

0:03:31.520000 --> 0:03:35.600000
 So as instances are added to the virtual
 machine scale set, those instances

0:03:35.600000 --> 0:03:38.960000
 are automatically properly registered
 with the load balancer.

0:03:38.960000 --> 0:03:42.820000
 And if instances are then removed from
 the virtual machine scale set,

0:03:42.820000 --> 0:03:46.180000
 that is also automatically updated
 with the load balancer.

0:03:46.180000 --> 0:03:54.160000
 The virtual machine scale set
 also supports auto scale.

0:03:54.160000 --> 0:04:03.560000
 Auto scale is actually a separate resource
 that monitors the virtual machine

0:04:03.560000 --> 0:04:10.160000
 scale set and can be set with rules to
 really manage the number of instances

0:04:10.160000 --> 0:04:14.980000
 that you have. The auto scale capability,
 the auto scale resource is actually

0:04:14.980000 --> 0:04:20.500000
 the same resource that's used for virtual
 machine scale sets and for app

0:04:20.500000 --> 0:04:23.960000
 services. So web apps and
 API apps, et cetera.

0:04:23.960000 --> 0:04:28.500000
 And it just monitors whatever criteria,
 the kind of the simplest one is

0:04:28.500000 --> 0:04:32.880000
 if you think about say percent of processor,
 but there's other criteria

0:04:32.880000 --> 0:04:39.140000
 as well. And it will add instances
 or it will take instances away.

0:04:39.140000 --> 0:04:42.460000
 Now another thing that you typically
 want with your virtual machine scale

0:04:42.460000 --> 0:04:47.240000
 set, as I add new instances, I do not
 want to have to go in and manually

0:04:47.240000 --> 0:04:49.400000
 configure a new instance.

0:04:49.400000 --> 0:04:52.900000
 I want each instance to come
 up and be ready to go.

0:04:52.900000 --> 0:04:54.760000
 There's a couple of different
 ways that you can do that.

0:04:54.760000 --> 0:05:00.620000
 One, I can base my instances
 on a custom image.

0:05:00.620000 --> 0:05:09.020000
 So I have a custom image here.

0:05:09.020000 --> 0:05:13.440000
 And it has all the software, all the
 configurations so that anytime a

0:05:13.440000 --> 0:05:19.560000
 new instance is provisioned, it automatically
 picks up everything it needs

0:05:19.560000 --> 0:05:21.880000
 from that custom image.

0:05:21.880000 --> 0:05:29.840000
 Or alternatively, I could use extensions
 such as the custom script extension.

0:05:29.840000 --> 0:05:40.800000
 And have that automatically fire up each
 time a new instance is provisioned.

0:05:40.800000 --> 0:05:44.580000
 So let's say for example, I've got a,
 I want to use a marketplace Ubuntu

0:05:44.580000 --> 0:05:50.740000
 image, but I've got a script that configures
 my own web service on that.

0:05:50.740000 --> 0:05:53.540000
 And I just want to make sure that that
 script gets run every time a new

0:05:53.540000 --> 0:05:57.340000
 instance is fired up, then I could
 set that up with a custom script.

0:05:57.340000 --> 0:06:01.420000
 Regardless of whether you go with some
 kind of extension for the automatic

0:06:01.420000 --> 0:06:05.420000
 configuration at the OS level, or if
 you use a custom image, you really

0:06:05.420000 --> 0:06:10.980000
 want to have a situation where you don't
 have to manually interact with

0:06:10.980000 --> 0:06:14.760000
 new instances. You can set it up, for
 example, through the load balancers

0:06:14.760000 --> 0:06:19.760000
 so that you have your NAT rules that
 would allow you to connect up and

0:06:19.760000 --> 0:06:24.560000
 establish SSH or establish RDP
 with individual instances.

0:06:24.560000 --> 0:06:28.680000
 And in fact, they're going
 to be on a virtual network.

0:06:28.680000 --> 0:06:31.300000
 If you can get to that virtual network,
 you can also connect to them.

0:06:31.300000 --> 0:06:34.440000
 But that's kind of antithetical
 to what you want to do.

0:06:34.440000 --> 0:06:38.260000
 My goal, if I have a virtual machine
 scale set, is that I should never

0:06:38.260000 --> 0:06:42.180000
 have to individually interact
 with the instances.

0:06:42.180000 --> 0:06:43.900000
 I want to interact
 with it as a whole.

0:06:43.900000 --> 0:06:48.660000
 Doesn't mean I can't in an emergency,
 but just conceptually I want that

0:06:48.660000 --> 0:06:52.160000
 to be an automated process because
 that's kind of the point of having

0:06:52.160000 --> 0:06:55.240000
 a scale set. All right, so that's
 the idea of a scale set.

0:06:55.240000 --> 0:06:59.000000
 Now let's talk about some
 of the concepts here.

0:06:59.000000 --> 0:07:04.520000
 First of all, as I mentioned, Azure
 AutoScale, the auto scale component

0:07:04.520000 --> 0:07:09.660000
 of a virtual machine scale set is managed
 by this independent resource,

0:07:09.660000 --> 0:07:11.940000
 if you will. It's
 obviously tied in.

0:07:11.940000 --> 0:07:14.780000
 And it applies to three things
 actually right now.

0:07:14.780000 --> 0:07:17.600000
 It applies, as I said, to the virtual
 machine scale set and to an app

0:07:17.600000 --> 0:07:21.500000
 service. You might see it applied to
 a cloud service, which is part of

0:07:21.500000 --> 0:07:25.640000
 what is called classic or the Azure service
 management model is the original

0:07:25.640000 --> 0:07:30.200000
 Azure model. They are still around,
 but anything new that you would do,

0:07:30.200000 --> 0:07:33.960000
 you would be doing as a virtual machine
 scale set or an app service.

0:07:33.960000 --> 0:07:38.580000
 So that's probably not something that's
 going to be really central to

0:07:38.580000 --> 0:07:40.160000
 the way you work in Azure.

0:07:40.160000 --> 0:07:45.500000
 Or, for example, if you're looking to
 study for the exam, the cloud service

0:07:45.500000 --> 0:07:49.320000
 would be extremely unlikely to be
 the right solution to anything.

0:07:49.320000 --> 0:07:52.320000
 But it is there if you see it, so
 you'll just know what that is.

0:07:52.320000 --> 0:07:56.920000
 What's interesting is that it is provided
 as a resource from a completely

0:07:56.920000 --> 0:07:59.780000
 different resource provider,
 still Microsoft.

0:07:59.780000 --> 0:08:03.260000
 Whereas a virtual machine scale set
 is provided through Microsoft.

0:08:03.260000 --> 0:08:07.480000
 Compute, the auto scale is actually
 provided through Microsoft.

0:08:07.480000 --> 0:08:13.820000
 Insights. And insights is actually just
 the original name of what is now

0:08:13.820000 --> 0:08:16.280000
 log analytics or that kind
 of whole framework.

0:08:16.280000 --> 0:08:19.020000
 It was kind of the foundational framework
 that log analytics has evolved

0:08:19.020000 --> 0:08:21.500000
 from. But that's neither
 here nor there.

0:08:21.500000 --> 0:08:24.100000
 But if you see that again, that's more
 one of those things that you just

0:08:24.100000 --> 0:08:27.080000
 know that's what you've got.

0:08:27.080000 --> 0:08:28.800000
 Now load balancing.

0:08:28.800000 --> 0:08:30.720000
 Couple things with load balancing.


0:08:30.720000 --> 0:08:34.600000
 First of all, you're going to integrate
 your virtual machine scale set

0:08:34.600000 --> 0:08:39.320000
 with either a load balancer or you
 can actually now integrate it with

0:08:39.320000 --> 0:08:41.000000
 an application gateway.

0:08:41.000000 --> 0:08:42.280000
 An application gateway.

0:08:42.280000 --> 0:08:44.440000
 I'm going to go with a load balancer
 just because honestly it's worth

0:08:44.440000 --> 0:08:46.100000
 telling it forever.

0:08:46.100000 --> 0:08:50.100000
 The load balancer, again, when it's
 integrated with a virtual machine

0:08:50.100000 --> 0:08:53.900000
 scale set, it's going to automatically
 account for scale operations.

0:08:53.900000 --> 0:08:58.860000
 You can set NAT rules for direct
 access to instances.

0:08:58.860000 --> 0:09:01.740000
 However, I don't recommend
 doing that.

0:09:01.740000 --> 0:09:04.780000
 Again, not that I don't
 recommend doing that.

0:09:04.780000 --> 0:09:06.480000
 I don't recommend relying on that.


0:09:06.480000 --> 0:09:09.380000
 Try to make sure that you
 never need to do that.

0:09:09.380000 --> 0:09:13.080000
 But maybe if you need to have it
 for emergency purposes, you can.

0:09:13.080000 --> 0:09:18.760000
 Also, you can deploy a virtual machine
 scale set from the portal.

0:09:18.760000 --> 0:09:21.580000
 Don't. It doesn't deploy
 quite right.

0:09:21.580000 --> 0:09:24.400000
 At least not at the time that
 I'm making this video.

0:09:24.400000 --> 0:09:28.720000
 This has really been the case ever
 since you could deploy through the

0:09:28.720000 --> 0:09:32.720000
 portal. What you want to do is you
 want to deploy via a template.

0:09:32.720000 --> 0:09:36.720000
 Now, in the demonstration, I am going
 to generate a template using a tool.

0:09:36.720000 --> 0:09:38.580000
 I'm actually going to
 use Visual Studio.

0:09:38.580000 --> 0:09:44.320000
 I usually use Visual Studio code, but
 Visual Studio actually has a little

0:09:44.320000 --> 0:09:48.460000
 bit better tooling, specifically for
 the virtual machine scale set.

0:09:48.460000 --> 0:09:54.400000
 However, also know that there are quite
 a number of examples of virtual

0:09:54.400000 --> 0:09:55.660000
 machine scale set templates.

0:09:55.660000 --> 0:10:00.120000
 In fact, there's a number of the Azure
 Quick Starts that give you variations

0:10:00.120000 --> 0:10:02.400000
 of virtual machine scale sets.

0:10:02.400000 --> 0:10:06.540000
 The reality is you will likely never
 have to create a virtual machine

0:10:06.540000 --> 0:10:08.560000
 scale set template from scratch.

0:10:08.560000 --> 0:10:12.380000
 Go pick one up that's close to what
 you need and then just modify it as

0:10:12.380000 --> 0:10:15.380000
 necessary. That is something you always
 want to do with virtual machine

0:10:15.380000 --> 0:10:19.840000
 scale set is deploy via a template because
 that's going to deploy it right.

0:10:19.840000 --> 0:10:23.520000
 It's just a little bit of
 a quirk with the system.

0:10:23.520000 --> 0:10:27.580000
 With that said, I am going to demonstrate
 a virtual machine scale set.

0:10:27.580000 --> 0:10:29.900000
 I'm going to provision a virtual
 machine scale set.

0:10:29.900000 --> 0:10:34.160000
 Again, I'm going to provision
 it via a template.

0:10:34.160000 --> 0:10:37.580000
 I'm going to build that
 template from scratch.

0:10:37.580000 --> 0:10:39.700000
 I'm actually not going to
 do a lot of the work.

0:10:39.700000 --> 0:10:42.280000
 The Visual Studio environment is
 going to do a lot of the work.

0:10:42.280000 --> 0:10:46.280000
 By the way, I'm using Visual
 Studio Express which is free.

0:10:46.280000 --> 0:10:49.640000
 The only thing with Visual Studio Express,
 unlike Visual Studio Code,

0:10:49.640000 --> 0:10:51.520000
 it's not cross-platform.

0:10:51.520000 --> 0:10:57.560000
 If you are working in Windows and you
 want a tool that has a lot of really

0:10:57.560000 --> 0:11:00.940000
 Azure related capabilities, including
 the ability to manage templates,

0:11:00.940000 --> 0:11:03.300000
 then you might want to
 look at Visual Studio.

0:11:03.300000 --> 0:11:06.660000
 I've kind of gotten to the point though
 that from a tooling standpoint,

0:11:06.660000 --> 0:11:11.900000
 almost always. I think I prefer
 Visual Studio Code.

0:11:11.900000 --> 0:11:14.460000
 It's kept, frankly, just
 kept up a little more.

0:11:14.460000 --> 0:11:16.220000
 It just doesn't have the virtual
 machine scale set.

0:11:16.220000 --> 0:11:19.800000
 Not a big deal. Hit a button I
 wasn't supposed to hit there.

0:11:19.800000 --> 0:11:24.580000
 All right. Let's go ahead and let's
 take a look at this demonstration.

0:11:24.580000 --> 0:11:29.580000
 Again, what I'm going to do is I'm
 going to generate a template for a

0:11:29.580000 --> 0:11:32.000000
 virtual machine scale
 set based on Linux.

0:11:32.000000 --> 0:11:40.520000
 Then I am also going to inject a custom
 script into that virtual machine

0:11:40.520000 --> 0:11:41.960000
 scale set template.

0:11:41.960000 --> 0:11:44.560000
 Now, what I'm doing is
 I'm in Visual Studio.

0:11:44.560000 --> 0:11:48.660000
 I've created what's called a Visual
 Studio Solution, which is a blank

0:11:48.660000 --> 0:11:53.960000
 solution. I'm going to
 add a project to this.

0:11:53.960000 --> 0:11:58.540000
 This is Visual Studio 2019.

0:11:58.540000 --> 0:12:02.420000
 I'm going to add an Azure
 Resource Group project.

0:12:02.420000 --> 0:12:10.440000
 I'm going to call
 this Deploy VMSS.

0:12:10.440000 --> 0:12:12.280000
 Framework doesn't matter.

0:12:12.280000 --> 0:12:17.580000
 It does matter is picking
 the right template.

0:12:17.580000 --> 0:12:22.020000
 Even here, even though I am quote unquote
 creating this from scratch,

0:12:22.020000 --> 0:12:26.080000
 I've got templates and I could also
 go and actually pull Quick Starts

0:12:26.080000 --> 0:12:28.620000
 directly, which is pretty cool.

0:12:28.620000 --> 0:12:30.380000
 That's what I want though.

0:12:30.380000 --> 0:12:31.820000
 Linux virtual machine scale set.

0:12:31.820000 --> 0:12:33.060000
 I'm going to click OK.

0:12:33.060000 --> 0:12:37.140000
 This is going to generate
 the Azure template for me.

0:12:37.140000 --> 0:12:45.100000
 Here is my Linux virtual machine
 scale set as a JSON file.

0:12:45.100000 --> 0:12:47.720000
 That out of the way.

0:12:47.720000 --> 0:12:51.000000
 I have a number of
 parameters here.

0:12:51.000000 --> 0:12:54.020000
 I've got the VM size.

0:12:54.020000 --> 0:13:06.500000
 I'm going to make that a D2S
 V3 because that is my go-to.

0:13:06.500000 --> 0:13:17.240000
 I'm going to make this 16.04 LTS.

0:13:17.240000 --> 0:13:22.600000
 That is kind of my go-to 16 or 18.


0:13:22.600000 --> 0:13:25.360000
 My go-to Ubuntu servers.

0:13:25.360000 --> 0:13:27.500000
 Name, we're good.

0:13:27.500000 --> 0:13:28.920000
 Instance count, I'm good.

0:13:28.920000 --> 0:13:31.760000
 Admin username, admin password,
 all of those.

0:13:31.760000 --> 0:13:38.660000
 I'm fine with those as parameters.


0:13:38.660000 --> 0:13:42.120000
 And locations. These
 are variables.

0:13:42.120000 --> 0:13:45.200000
 Now I'm going to make some changes to
 this because for some reason this

0:13:45.200000 --> 0:13:51.460000
 still does this with unmanaged disks
 and I'm going to change that.

0:13:51.460000 --> 0:13:56.740000
 We're going to leave all of this in
 here because I'm not going to trim

0:13:56.740000 --> 0:14:01.580000
 it down as I normally would.

0:14:01.580000 --> 0:14:03.200000
 And that's all good.

0:14:03.200000 --> 0:14:05.800000
 And so my virtual network,
 I've got a virtual network.

0:14:05.800000 --> 0:14:08.580000
 It's got an address,
 it's got a subnet.

0:14:08.580000 --> 0:14:16.000000
 I've got a storage account that's getting
 generated based on this array

0:14:16.000000 --> 0:14:18.460000
 of unique strings.

0:14:18.460000 --> 0:14:22.620000
 And I'll let it create
 all of them.

0:14:22.620000 --> 0:14:28.280000
 Got a public IP address that is going to
 be associated with my load balancer.

0:14:28.280000 --> 0:14:35.100000
 I have a load balancer and the load balancer
 has a NAT rule which is going

0:14:35.100000 --> 0:14:40.440000
 to bring in the front end 50,000 to
 50,000 in nine, mapping that to the

0:14:40.440000 --> 0:14:44.320000
 back import of 22.

0:14:44.320000 --> 0:14:49.280000
 And come down here and I've got
 my virtual machine scale set.

0:14:49.280000 --> 0:14:59.180000
 Now, if I want, I could
 go 26, 2017, 03,30.

0:14:59.180000 --> 0:15:07.960000
 Give me an API version that
 lets me use managed disks.

0:15:07.960000 --> 0:15:17.420000
 And we're going to really pretty
 much take everything.

0:15:17.420000 --> 0:15:28.140000
 And what I'm going to do is delete
 that and that should be good.

0:15:28.140000 --> 0:15:32.700000
 All right. And computer name prefix, fine
 admin username, password, network

0:15:32.700000 --> 0:15:35.740000
 profile. Now the network profile for a
 virtual machine scale set is different

0:15:35.740000 --> 0:15:41.500000
 than the network profile for a virtual
 machine because rather than tying

0:15:41.500000 --> 0:15:48.420000
 to an existing NIC, you're actually
 going to define the entire network

0:15:48.420000 --> 0:15:52.500000
 interface configuration
 in the network profile.

0:15:52.500000 --> 0:15:57.680000
 And then it's got a diagnostics extension
 which is quite well extensive.

0:15:57.680000 --> 0:16:01.560000
 All right. So this is actually, you
 can see it's 257 lines until we get

0:16:01.560000 --> 0:16:06.780000
 there. And then it is a good
 bit more from 257 to 519.

0:16:06.780000 --> 0:16:10.300000
 It's all just the diagnostics
 extension.

0:16:10.300000 --> 0:16:12.920000
 Now, in addition to that, I'm going
 to add a few more things.

0:16:12.920000 --> 0:16:18.800000
 I'm going to go over here to my outline
 and I'm going to add a new resource.

0:16:18.800000 --> 0:16:23.540000
 And what I want to do is add
 an auto scale resource.

0:16:23.540000 --> 0:16:30.720000
 And this is going to be demo auto scale
 because I have no creativity.

0:16:30.720000 --> 0:16:32.740000
 I'm simply going to add that.

0:16:32.740000 --> 0:16:40.380000
 And what that does is it
 sets up auto scale rules.

0:16:40.380000 --> 0:16:46.660000
 And my auto scale rules, I'm
 using this over, there we go.

0:16:46.660000 --> 0:16:50.000000
 I see that the auto scale rules come
 from Microsoft Insights, auto scale

0:16:50.000000 --> 0:16:51.540000
 settings as I said.

0:16:51.540000 --> 0:16:54.780000
 It, of course, depends on the
 virtual machine scale set.

0:16:54.780000 --> 0:16:59.180000
 And it goes down and
 just sets some rules.

0:16:59.180000 --> 0:17:00.980000
 So this goes between one and ten.

0:17:00.980000 --> 0:17:03.620000
 We're going to save between
 one and three.

0:17:03.620000 --> 0:17:07.400000
 And then it's got metrics.

0:17:07.400000 --> 0:17:10.740000
 And so it's looking for the processor,
 which is a variable.

0:17:10.740000 --> 0:17:13.860000
 And it's just looking, if it's greater
 than 80, it's going to scale up.

0:17:13.860000 --> 0:17:16.940000
 And if it's under, I think what is
 it, 60, it's going to scale down.

0:17:16.940000 --> 0:17:20.060000
 There's just basic rules that
 it's putting in place.

0:17:20.060000 --> 0:17:21.500000
 And that's fine.

0:17:21.500000 --> 0:17:28.400000
 And then I want to do one other thing,
 which is I'm going to add a custom

0:17:28.400000 --> 0:17:32.140000
 script for Linux VMSs.

0:17:32.140000 --> 0:17:42.100000
 And we'll say demo scripts.

0:17:42.100000 --> 0:17:45.460000
 Apparently it didn't want
 to do demo script.

0:17:45.460000 --> 0:17:55.980000
 Let's try that again.

0:17:55.980000 --> 0:17:59.080000
 And I hit cancel instead of add.

0:17:59.080000 --> 0:18:01.960000
 All right. Now I did.

0:18:01.960000 --> 0:18:02.840000
 And it just didn't pick it up.

0:18:02.840000 --> 0:18:04.440000
 That's intriguing.

0:18:04.440000 --> 0:18:14.520000
 One moment. Delete that.

0:18:14.520000 --> 0:18:15.760000
 And we'll try this again.

0:18:15.760000 --> 0:18:18.200000
 Hadn't had that one
 happen to me before.

0:18:18.200000 --> 0:18:29.740000
 I really wanted to show you how you can
 add a script to a virtual machine

0:18:29.740000 --> 0:18:33.580000
 scale set. All right.

0:18:33.580000 --> 0:18:37.200000
 Well, this does not appear
 to want to add it in.

0:18:37.200000 --> 0:18:41.780000
 That is an order added in twice.

0:18:41.780000 --> 0:18:45.060000
 All right. We're going to pause here
 for a moment while I figure out why

0:18:45.060000 --> 0:18:46.720000
 it's, oh, there it is.

0:18:46.720000 --> 0:18:47.580000
 It did add it in twice.

0:18:47.580000 --> 0:18:49.060000
 We're not going to pause.

0:18:49.060000 --> 0:18:51.700000
 It just was lying to me.

0:18:51.700000 --> 0:18:54.080000
 Okay. There we go.

0:18:54.080000 --> 0:18:54.760000
 We'll take that out.

0:18:54.760000 --> 0:18:56.820000
 So it's only going in there once.

0:18:56.820000 --> 0:18:59.740000
 And now it has this file URIs.

0:18:59.740000 --> 0:19:03.820000
 And normally what it does, what this
 tool does, is it actually generates

0:19:03.820000 --> 0:19:06.560000
 a script file for you that you could
 go in and you could set up.

0:19:06.560000 --> 0:19:10.160000
 But I already have a script file.

0:19:10.160000 --> 0:19:15.860000
 So what I want to do is I want to
 actually pick up that script file.

0:19:15.860000 --> 0:19:21.380000
 Or at least pick up the URL
 for that script file.

0:19:21.380000 --> 0:19:26.880000
 That out of the way.

0:19:26.880000 --> 0:19:32.120000
 And replace file UR with
 an existing script file.

0:19:32.120000 --> 0:19:37.320000
 And this is just a file that installs
 a very simple web server.

0:19:37.320000 --> 0:19:40.800000
 It's just a Python
 script web server.

0:19:40.800000 --> 0:19:52.500000
 And then I want that to run
 just this bit right here.

0:19:52.500000 --> 0:20:02.420000
 All right. And I have now
 set up a fairly complex.

0:20:02.420000 --> 0:20:05.480000
 Oh, let's see. Are we good there?

0:20:05.480000 --> 0:20:08.900000
 Nope. I didn't delete
 the demo script.

0:20:08.900000 --> 0:20:12.320000
 Oh, I deleted one.

0:20:12.320000 --> 0:20:13.780000
 So I'd done that like three times.


0:20:13.780000 --> 0:20:15.700000
 There we go. Now we're good.

0:20:15.700000 --> 0:20:17.920000
 All right. Third time's charm.

0:20:17.920000 --> 0:20:20.340000
 All right. But I've got a demo
 script that I've added.

0:20:20.340000 --> 0:20:29.080000
 It already added the ability
 to capture diagnostic data.

0:20:29.080000 --> 0:20:30.740000
 I've added auto scale.

0:20:30.740000 --> 0:20:34.680000
 And so I've got everything that I want
 for this particular deployment.

0:20:34.680000 --> 0:20:38.000000
 And this tool actually lets me
 deploy directly from here.

0:20:38.000000 --> 0:20:40.620000
 So I'm going to go
 ahead and deploy.

0:20:40.620000 --> 0:20:44.320000
 I have to wait for this to come
 up with my subscription.

0:20:44.320000 --> 0:20:47.080000
 There we go. That is the
 subscription that I want.

0:20:47.080000 --> 0:20:49.300000
 Create a new deployment.

0:20:49.300000 --> 0:20:59.220000
 Putting it as always
 in the east us.

0:20:59.220000 --> 0:21:06.260000
 Now the next thing I'm going to do,
 it actually has a template parameter

0:21:06.260000 --> 0:21:09.460000
 file. I'm going to go ahead
 and edit the parameters.

0:21:09.460000 --> 0:21:12.660000
 And we're going to go in
 and give this a name of.

0:21:12.660000 --> 0:21:18.740000
 I need a demo of VMSS.

0:21:18.740000 --> 0:21:21.620000
 I want two instances.

0:21:21.620000 --> 0:21:24.420000
 Admin username of student.

0:21:24.420000 --> 0:21:26.640000
 I'm not going to put in
 the admin password.

0:21:26.640000 --> 0:21:29.960000
 Now what I would normally do, actually
 I'm going to do this just to show

0:21:29.960000 --> 0:21:32.860000
 you. Totally off topic,
 but pretty cool.

0:21:32.860000 --> 0:21:35.260000
 I don't want to put a sensitive
 piece of data in here.

0:21:35.260000 --> 0:21:43.780000
 I've actually got a key vault setup
 where I have an admin password in

0:21:43.780000 --> 0:21:47.800000
 my key vault. And this tool actually
 let me go and just pick that up.

0:21:47.800000 --> 0:21:54.980000
 And it's going to inject the right markup
 so that I pull my password out

0:21:54.980000 --> 0:21:57.620000
 of that key vault.

0:21:57.620000 --> 0:22:00.640000
 So the password is never
 in plain text.

0:22:00.640000 --> 0:22:02.900000
 And then I deploy.

0:22:02.900000 --> 0:22:05.860000
 And once I start deploying, I'm going
 to wait for a little while.

0:22:05.860000 --> 0:22:08.380000
 It's going to take a few minutes to
 come back and we'll see if we have

0:22:08.380000 --> 0:22:09.980000
 a virtual machine scale set.

0:22:09.980000 --> 0:22:17.840000
 All right, if we look at this, we can
 see that this was a swimming success.

0:22:17.840000 --> 0:22:19.060000
 Look here, it says succeeded.

0:22:19.060000 --> 0:22:23.580000
 Now I'm going to pull back the curtain
 and be fully forthright with you

0:22:23.580000 --> 0:22:26.500000
 because otherwise would
 not be a good thing.

0:22:26.500000 --> 0:22:31.000000
 It's entirely possible that this failed
 the first time and then was very

0:22:31.000000 --> 0:22:34.360000
 stubborn about me getting it fixed
 until such time as I realized what

0:22:34.360000 --> 0:22:39.080000
 was happening. And I think it's a worthwhile
 exercise for you to see what

0:22:39.080000 --> 0:22:44.840000
 actually did happen and why it took a
 while for that to actually complete

0:22:44.840000 --> 0:22:48.480000
 successfully. All right, if I go down
 here to my definition of the virtual

0:22:48.480000 --> 0:22:55.020000
 machine scale set and if I go to, let's
 collapse a bunch of stuff, if

0:22:55.020000 --> 0:22:59.880000
 I go to my script where I was deploying
 the script, when I first deployed

0:22:59.880000 --> 0:23:02.620000
 it, I forgot to put
 the SH in there.

0:23:02.620000 --> 0:23:06.160000
 And so when it deployed and it tried
 to deploy the extension, I got an

0:23:06.160000 --> 0:23:09.060000
 error. And then I fixed it, I tried
 to deploy again, I kept getting the

0:23:09.060000 --> 0:23:16.120000
 same error. I was getting the same error
 because even though the actual

0:23:16.120000 --> 0:23:19.400000
 template and deployment of template
 is what we call a dampenant and that

0:23:19.400000 --> 0:23:23.720000
 you can redeploy it and it will be fine,
 because of the way that extensions

0:23:23.720000 --> 0:23:28.060000
 work, that extension was still there
 and it just kept trying to rerun

0:23:28.060000 --> 0:23:30.260000
 the same extension that was
 already set up incorrectly.

0:23:30.260000 --> 0:23:33.900000
 So I just went and I deployed it after
 this fix, I deployed it somewhere

0:23:33.900000 --> 0:23:37.220000
 else. And so now there's a slightly
 different name to the resource group

0:23:37.220000 --> 0:23:39.360000
 and to the actual virtual
 machine scale set.

0:23:39.360000 --> 0:23:43.320000
 But that was the only thing that changed
 from when I walked through it

0:23:43.320000 --> 0:23:44.840000
 initially and now.

0:23:44.840000 --> 0:23:46.140000
 But I think that's worthwhile.

0:23:46.140000 --> 0:23:47.440000
 I have to make sure
 you get that right.

0:23:47.440000 --> 0:23:51.340000
 Also with that redeployability of templates
 when things don't go right,

0:23:51.340000 --> 0:23:54.000000
 this is one of those little cases
 where it's not exactly the case.

0:23:54.000000 --> 0:23:56.780000
 What I could have done was I could
 have actually gone into the virtual

0:23:56.780000 --> 0:24:02.260000
 machine scale set and deleted that failed
 extension and then redeployed

0:24:02.260000 --> 0:24:03.680000
 it would have been fine.

0:24:03.680000 --> 0:24:06.780000
 But that actually would have taken longer
 than just redeploying the whole

0:24:06.780000 --> 0:24:08.460000
 thing. So that's what I did.

0:24:08.460000 --> 0:24:10.880000
 And it has now deployed properly.

0:24:10.880000 --> 0:24:15.660000
 So what I want to do is take you over
 to the resource group that this

0:24:15.660000 --> 0:24:17.040000
 all deployed into.

0:24:17.040000 --> 0:24:22.460000
 So here's my resource group and I've
 got some, really the majority of

0:24:22.460000 --> 0:24:24.440000
 what you see here are
 the storage accounts.

0:24:24.440000 --> 0:24:27.320000
 Now I will tell you, by and large, these
 storage accounts are not actually

0:24:27.320000 --> 0:24:34.040000
 necessary because I made this
 a managed disk system.

0:24:34.040000 --> 0:24:38.360000
 But one of these holds the diagnostic
 data that's used for auto scale.

0:24:38.360000 --> 0:24:40.520000
 So I just let it run as is, right?


0:24:40.520000 --> 0:24:44.440000
 But I could have actually eliminated
 that had one storage account, one

0:24:44.440000 --> 0:24:47.900000
 virtual machine scale set, load balancer,
 public IP address and virtual

0:24:47.900000 --> 0:24:52.960000
 network. And I'm going to take a look
 at my virtual machine scale set.

0:24:52.960000 --> 0:24:57.280000
 And there's generally not a lot that
 you're going to do with this.

0:24:57.280000 --> 0:25:00.320000
 Right now I've got one
 instance running.

0:25:00.320000 --> 0:25:04.220000
 Easy enough. Now I could
 go into the instance.

0:25:04.220000 --> 0:25:06.980000
 You actually can get some
 instance information.

0:25:06.980000 --> 0:25:11.600000
 So there is some drill down capability,
 including connecting to it via

0:25:11.600000 --> 0:25:13.980000
 serial console using the bastion.

0:25:13.980000 --> 0:25:17.700000
 If you have that set up or going
 in with boot diagnostics.

0:25:17.700000 --> 0:25:19.440000
 But I don't want to do
 any of that again.

0:25:19.440000 --> 0:25:23.840000
 I want to kind of stay away
 from my instance management.

0:25:23.840000 --> 0:25:25.400000
 I can go to my scaling.

0:25:25.400000 --> 0:25:27.980000
 This is where I set up
 my auto scale rule.

0:25:27.980000 --> 0:25:31.420000
 Right? You can see it's got a scale up
 when it's over 80, scale down over

0:25:31.420000 --> 0:25:38.820000
 60. And if I go to run history, it
 actually auto scaled down, which is

0:25:38.820000 --> 0:25:41.320000
 kind of interesting because
 it was just deployed.

0:25:41.320000 --> 0:25:44.960000
 But it auto scaled down because my setting
 for the virtual machine scale

0:25:44.960000 --> 0:25:47.420000
 set was to have two instances.

0:25:47.420000 --> 0:25:51.360000
 But the auto scale actually sets
 up a default of one instance.

0:25:51.360000 --> 0:25:53.360000
 So I did that intentionally.

0:25:53.360000 --> 0:25:54.720000
 I'm not going to go through
 scaling this.

0:25:54.720000 --> 0:25:57.440000
 But you can see where that
 auto scale does kick in.

0:25:57.440000 --> 0:26:00.440000
 It does override other settings,
 which is pretty cool.

0:26:00.440000 --> 0:26:05.360000
 And other things that you can set
 up, I can actually add data disks.

0:26:05.360000 --> 0:26:07.940000
 And that's kind of interesting.

0:26:07.940000 --> 0:26:13.780000
 Just some settings in terms of your
 computer name prefix, administrator,

0:26:13.780000 --> 0:26:18.900000
 et cetera. And you can change the size
 and just a number of other things

0:26:18.900000 --> 0:26:21.380000
 you can do. But we do have
 this thing deployed.

0:26:21.380000 --> 0:26:25.180000
 Now, I did also, since I kind of had
 to take break anyways and fix that,

0:26:25.180000 --> 0:26:29.100000
 I did modify the load balancer.

0:26:29.100000 --> 0:26:31.900000
 So the load balancer came in
 with an inbound NAT rule.

0:26:31.900000 --> 0:26:33.880000
 I am going to show
 you the NAT rule.

0:26:33.880000 --> 0:26:40.160000
 This sets up so that incoming NAT
 is going to go over 50,000.

0:26:40.160000 --> 0:26:41.300000
 We'll go to the first instance.

0:26:41.300000 --> 0:26:45.020000
 If I change the incoming IP to 50,000
 in one, it would go to the second

0:26:45.020000 --> 0:26:48.880000
 instance. Now, how does
 that actually work?

0:26:48.880000 --> 0:26:50.660000
 It's actually pretty cool.

0:26:50.660000 --> 0:26:53.300000
 I'm going to pull over.

0:26:53.300000 --> 0:26:55.140000
 I'm in Windows 10.

0:26:55.140000 --> 0:26:57.160000
 And one of the nice things with Windows
 10 now is it actually has a built

0:26:57.160000 --> 0:27:01.220000
-in SSH client. And
 I can go to SSH.

0:27:01.220000 --> 0:27:07.180000
 And then student is the name of the
 admin account that I gave in there.

0:27:07.180000 --> 0:27:12.280000
 And then this is actually the DNS name
 that I associated with the load

0:27:12.280000 --> 0:27:16.260000
 balancer. And I'm going to say that
 I'm going to connect over 40,000.

0:27:16.260000 --> 0:27:20.200000
 And when I do that, it's going to ask
 me for the password that I used

0:27:20.200000 --> 0:27:24.080000
 that's going to be the same password
 on all of the machines.

0:27:24.080000 --> 0:27:26.780000
 And I am now into one
 of the instances.

0:27:26.780000 --> 0:27:28.700000
 In fact, I'm into
 the only instance.

0:27:28.700000 --> 0:27:30.160000
 Now, I can do things, for example.


0:27:30.160000 --> 0:27:32.700000
 I could go, should be here.

0:27:32.700000 --> 0:27:35.280000
 P.S. Ox, I'm going to grep web.

0:27:35.280000 --> 0:27:36.380000
 That's my web server.

0:27:36.380000 --> 0:27:39.060000
 I can see that the web
 server is running.

0:27:39.060000 --> 0:27:40.260000
 And that's that first line.

0:27:40.260000 --> 0:27:42.600000
 If you're not a Linux person,
 it doesn't really matter.

0:27:42.600000 --> 0:27:43.980000
 So that means that cool.

0:27:43.980000 --> 0:27:45.700000
 It is actually running that.

0:27:45.700000 --> 0:27:53.900000
 It is scalable. And if I go to my public
 IP address, and if I grab the

0:27:53.900000 --> 0:27:56.480000
 DNS name, I could also
 use the IP address.

0:27:56.480000 --> 0:27:58.780000
 But since I gave it a DNS name.

0:27:58.780000 --> 0:28:05.440000
 And if I come over here and run this,
 it should come up with a little

0:28:05.440000 --> 0:28:07.900000
 bit. There's a moment or two.

0:28:07.900000 --> 0:28:11.020000
 I know it did when I
 ran this previously.

0:28:11.020000 --> 0:28:15.700000
 It's an extremely not
 well defined web app.

0:28:15.700000 --> 0:28:19.220000
 And so sometimes it gets
 a little bit confused.

0:28:19.220000 --> 0:28:24.120000
 So what I'm going to do here
 is pretend like it just ran.

0:28:24.120000 --> 0:28:27.380000
 That I ran, it's the same URL.

0:28:27.380000 --> 0:28:32.200000
 And what you see there is actually
 the instance name that it's running

0:28:32.200000 --> 0:28:36.620000
 on. And as I get new instances, of
 course, those numbers are going to

0:28:36.620000 --> 0:28:39.760000
 grow. Now, I want to show
 you something else.

0:28:39.760000 --> 0:28:40.460000
 Hey, there you go.

0:28:40.460000 --> 0:28:42.160000
 It took a while, but
 it did come up.

0:28:42.160000 --> 0:28:45.100000
 I want to show you something else with
 virtual machine scale sets because

0:28:45.100000 --> 0:28:46.900000
 I didn't really talk
 about the limits.

0:28:46.900000 --> 0:28:53.780000
 So I'm going to go
 azure, VMSS limits.

0:28:53.780000 --> 0:28:58.500000
 And I'm actually going to go ahead and
 go to the azure subscription limits.

0:28:58.500000 --> 0:29:00.960000
 First, we'll pop that one up.

0:29:00.960000 --> 0:29:04.360000
 And I'm also going to pop
 up the scale set overview.

0:29:04.360000 --> 0:29:11.260000
 I'll go here because I don't think this
 one has really great info on the

0:29:11.260000 --> 0:29:13.780000
 virtual machine scale
 set if I recall.

0:29:13.780000 --> 0:29:16.860000
 All right. Well, there we go.

0:29:16.860000 --> 0:29:17.780000
 It does, in fact.

0:29:17.780000 --> 0:29:20.780000
 Maximum number of VMs
 in a scale set.

0:29:20.780000 --> 0:29:26.540000
 You can have up to 1,000
 VMs per scale set.

0:29:26.540000 --> 0:29:30.040000
 Maximum number of VMs based on a custom
 image, but you can only go up

0:29:30.040000 --> 0:29:34.280000
 to 600 if you're using
 a custom image.

0:29:34.280000 --> 0:29:38.840000
 Now, you can also have up to
 2,000 scale sets in a region.

0:29:38.840000 --> 0:29:43.100000
 So if you were to combine this with some
 load balancing, then you're talking

0:29:43.100000 --> 0:29:48.300000
 about conceptually up to 2 million instances
 that could be running a workload

0:29:48.300000 --> 0:29:50.840000
 in a region through scale sets.

0:29:50.840000 --> 0:29:53.680000
 So you are talking about something
 that has the potential for massive

0:29:53.680000 --> 0:29:58.100000
 scale. I did want to show you that
 there is a difference in the number

0:29:58.100000 --> 0:30:03.660000
 of instances that you have, whether
 you're using a marketplace image or

0:30:03.660000 --> 0:30:07.780000
 a custom image. And so this would suggest
 that a marketplace image would

0:30:07.780000 --> 0:30:09.120000
 be the better solution.

0:30:09.120000 --> 0:30:14.180000
 However, keep in mind when you use a
 marketplace image, you're also going

0:30:14.180000 --> 0:30:18.480000
 to have to inject whatever that configuration
 is through some extension.

0:30:18.480000 --> 0:30:21.380000
 And that's going to be added
 time on scale operations.

0:30:21.380000 --> 0:30:26.700000
 And in general, keep in mind that when
 you auto scale, that's not an instance.

0:30:26.700000 --> 0:30:30.400000
 It's going to provision a new
 underlying virtual machine.

0:30:30.400000 --> 0:30:31.660000
 It's going to configure it.

0:30:31.660000 --> 0:30:33.320000
 And then it's going
 to become available.

0:30:33.320000 --> 0:30:37.620000
 So understand that auto scale with
 a virtual machine scale set is not

0:30:37.620000 --> 0:30:39.780000
 an instance. It's not an instant.

0:30:39.780000 --> 0:30:44.440000
 You want to optimize the load time for
 your virtual machine so that your

0:30:44.440000 --> 0:30:47.020000
 scale set can be as responsive
 as possible.

0:30:47.020000 --> 0:30:51.700000
 But in any case, that's really the concept
 of virtual machine scale sets.

0:30:51.700000 --> 0:30:55.060000
 You've seen what they are, how they
 work, what their components are, and

0:30:55.060000 --> 0:30:57.980000
 an example of a virtual machine
 scale set in action.

