1 00:00:00,540 --> 00:00:02,560 Hyper‑V high availability. 2 00:00:02,560 --> 00:00:04,670 Let's start by defining our terms. 3 00:00:04,670 --> 00:00:06,980 What exactly is meant by high availability? 4 00:00:06,980 --> 00:00:09,710 High availability for our context, 5 00:00:09,710 --> 00:00:13,140 and you can see the my citation in the lower part of the slide, 6 00:00:13,140 --> 00:00:17,640 involves the implementation of controls that guarantee your workload is up 7 00:00:17,640 --> 00:00:20,860 and running despite maintenance operations and outages. 8 00:00:20,860 --> 00:00:21,130 All right? 9 00:00:21,130 --> 00:00:24,950 Now, the words on this definition I think are very intentional. 10 00:00:24,950 --> 00:00:26,130 We have guarantee. 11 00:00:26,130 --> 00:00:30,040 You may very well be subject to service‑level agreements, 12 00:00:30,040 --> 00:00:33,280 SLAs, with your employees, with your colleagues, 13 00:00:33,280 --> 00:00:34,950 with your customers, whomever, 14 00:00:34,950 --> 00:00:39,700 where you actually quantify the degree of availability your workload has. 15 00:00:39,700 --> 00:00:44,130 And if your service is down over that SLA amount due 16 00:00:44,130 --> 00:00:46,160 to a planned or unplanned event, 17 00:00:46,160 --> 00:00:48,950 this could have big repercussions for your business, 18 00:00:48,950 --> 00:00:51,220 and frankly, for your job security. 19 00:00:51,220 --> 00:00:54,460 So we need to understand in Hyper‑V, and speaking as 20 00:00:54,460 --> 00:00:56,730 Hyper‑V hybrid cloud administrators, 21 00:00:56,730 --> 00:01:00,400 what are the main Microsoft‑provided options for doing high 22 00:01:00,400 --> 00:01:04,140 availability workloads in Hyper‑V virtual machines? 23 00:01:04,140 --> 00:01:06,640 Well, there's two kinds of clustering we'll learn a bit about. 24 00:01:06,640 --> 00:01:10,560 There's host clustering and guest clustering, and we also have the 25 00:01:10,560 --> 00:01:13,350 Windows Server Network Load Balancing service, 26 00:01:13,350 --> 00:01:17,940 also called NLB. Now I'm not going to get deep into the weeds 27 00:01:17,940 --> 00:01:23,160 in Windows Server failover clustering here because exam AZ‑801 28 00:01:23,160 --> 00:01:25,260 goes much deeper into clustering. 29 00:01:25,260 --> 00:01:28,530 So at this point, I just want to be able to differentiate, or I 30 00:01:28,530 --> 00:01:31,630 want you to be able to differentiate, the two types of 31 00:01:31,630 --> 00:01:33,540 clustering that we're talking about here. 32 00:01:33,540 --> 00:01:33,690 Now, 33 00:01:33,690 --> 00:01:37,760 first of all, to take a step back, clustering is generally speaking 34 00:01:37,760 --> 00:01:42,600 where you have shared storage between N number of nodes 35 00:01:42,600 --> 00:01:46,470 participating in the cluster. Now I say generally speaking because 36 00:01:46,470 --> 00:01:50,080 there are technologies where you can do Windows Server failover 37 00:01:50,080 --> 00:01:52,150 clustering without shared storage. 38 00:01:52,150 --> 00:01:56,220 We'll get into that in the AZ‑801 skill set, but let's just kick 39 00:01:56,220 --> 00:01:59,670 it old school and assume that we've got network accessible or 40 00:01:59,670 --> 00:02:03,200 locally attached shared storage and we have N number of nodes 41 00:02:03,200 --> 00:02:04,830 participating in the cluster. 42 00:02:04,830 --> 00:02:05,020 Now, 43 00:02:05,020 --> 00:02:08,140 the way that Windows Server failover clustering works is that 44 00:02:08,140 --> 00:02:11,930 one node is going to be the active node for that clustered 45 00:02:11,930 --> 00:02:14,000 role, and one of the most common, 46 00:02:14,000 --> 00:02:18,430 if not the most common clustered role, is the Hyper‑V virtual machine. 47 00:02:18,430 --> 00:02:20,800 So in this case, the right‑most node, 48 00:02:20,800 --> 00:02:24,120 the one that's in the three o'clock position, is active for that 49 00:02:24,120 --> 00:02:27,980 VM, and connectivity happens to that VM by coming into the 50 00:02:27,980 --> 00:02:30,980 failover cluster's virtual IP address. 51 00:02:30,980 --> 00:02:31,250 Now, 52 00:02:31,250 --> 00:02:35,540 what makes failover clustering a high availability solution is 53 00:02:35,540 --> 00:02:38,550 that if we need to take that node offline, 54 00:02:38,550 --> 00:02:44,110 the one that's hosting the VM, we can manually move that VM to another node in 55 00:02:44,110 --> 00:02:48,430 the cluster with minimal or preferably no service disruption. 56 00:02:48,430 --> 00:02:49,220 Likewise, 57 00:02:49,220 --> 00:02:53,610 if that node were to go down unexpectedly, the other nodes in the cluster 58 00:02:53,610 --> 00:02:58,530 are aware via heartbeat messages who's up and who's down, and in that case, 59 00:02:58,530 --> 00:03:01,470 another node will grab control over that VM. 60 00:03:01,470 --> 00:03:04,500 How is the node able to grasp that control? 61 00:03:04,500 --> 00:03:04,720 Well, 62 00:03:04,720 --> 00:03:09,100 because the VM and its definitions are in that shared storage that's 63 00:03:09,100 --> 00:03:13,070 accessible to all of the nodes in the cluster. You see the magic there? The 64 00:03:13,070 --> 00:03:17,670 reason why we're calling this Hyper‑V host clustering is because, in this 65 00:03:17,670 --> 00:03:23,020 case, your three nodes that are in this diagram are Hyper‑V hosts, as well 66 00:03:23,020 --> 00:03:27,330 as members of a failover cluster, and the virtual machine is what's called 67 00:03:27,330 --> 00:03:28,420 a clustered role. 68 00:03:28,420 --> 00:03:33,510 Now, there is a more specialized version of Windows Server failover clustering. 69 00:03:33,510 --> 00:03:37,640 What if you have a virtualized infrastructure where you have your 70 00:03:37,640 --> 00:03:43,150 Hyper‑V hosts, in this case, NodeA1, A2, and A3, and they are, of 71 00:03:43,150 --> 00:03:47,050 course, hosting virtual machines and highly available file shares and 72 00:03:47,050 --> 00:03:51,420 SQL Server databases, etc., and offering those clustered roles under 73 00:03:51,420 --> 00:03:53,310 virtual IPs as usual. 74 00:03:53,310 --> 00:03:58,850 But what if you wanted to create an internal failover cluster within your 75 00:03:58,850 --> 00:04:02,740 virtual machine infrastructure? A‑ha! In other words, 76 00:04:02,740 --> 00:04:07,240 from within the guest VMs. Is it possible to set up a failover 77 00:04:07,240 --> 00:04:10,870 cluster and advertise services under a virtual IP? 78 00:04:10,870 --> 00:04:13,000 And, of course, the answer to that is yes. This is 79 00:04:13,000 --> 00:04:15,840 called Hyper‑V guest clustering. 80 00:04:15,840 --> 00:04:20,490 Now, there are quite a few implications in terms of how you'd want this to work. 81 00:04:20,490 --> 00:04:24,030 Of course, one big one would be your cluster nodes really 82 00:04:24,030 --> 00:04:26,810 should be on top‑level cluster nodes. 83 00:04:26,810 --> 00:04:27,360 In other words, 84 00:04:27,360 --> 00:04:30,740 notice that I've gotten NodeB1 virtual machine that's 85 00:04:30,740 --> 00:04:32,800 being managed by the active node, 86 00:04:32,800 --> 00:04:39,830 NodeA1, but NodeB2, another member of that guest cluster, is on NodeA2. 87 00:04:39,830 --> 00:04:44,230 So what you have here is two levels of high availability to where you're 88 00:04:44,230 --> 00:04:48,830 protecting the VMs that are forming the guest cluster, and then within 89 00:04:48,830 --> 00:04:51,450 the guest cluster VM's operating system, 90 00:04:51,450 --> 00:04:54,650 there's another failover cluster where you've got high 91 00:04:54,650 --> 00:05:00,000 availability for services there. Yes, it's kind of complex and meta, but there you go.