1 00:00:00,740 --> 00:00:01,370 To that point, 2 00:00:01,370 --> 00:00:04,760 our goal here is to understand Storage Spaces and 3 00:00:04,760 --> 00:00:06,910 Storage Replica in Windows Server. 4 00:00:06,910 --> 00:00:07,900 Here's the use case. 5 00:00:07,900 --> 00:00:10,740 Dynamic disks and Windows Server deprecated, 6 00:00:10,740 --> 00:00:14,710 but we still want to do more advanced operations, 7 00:00:14,710 --> 00:00:19,410 like being able to expand a volume across physical disks and set up 8 00:00:19,410 --> 00:00:22,920 mirroring and parity and hot spares and that kind of stuff. 9 00:00:22,920 --> 00:00:26,380 In the absence of dynamic disks, what is available? 10 00:00:26,380 --> 00:00:29,100 Well, the answer to that question is Storage Spaces. 11 00:00:29,100 --> 00:00:32,650 And if you recognize that floppy disk icon, you're old like me. 12 00:00:32,650 --> 00:00:33,180 Congratulations! 13 00:00:33,180 --> 00:00:38,320 Storage Spaces is a storage virtualization and volume 14 00:00:38,320 --> 00:00:41,340 management technology built into Windows Server. 15 00:00:41,340 --> 00:00:44,400 It has, as its great advantage, no additional cost. 16 00:00:44,400 --> 00:00:46,530 You don't have to buy extra hardware. 17 00:00:46,530 --> 00:00:47,070 Now, of course, 18 00:00:47,070 --> 00:00:51,620 there's a tradeoff always when you're doing some kind of operation 19 00:00:51,620 --> 00:00:54,510 in software as opposed to dedicated hardware. 20 00:00:54,510 --> 00:00:56,250 But this is what we're dealing with. 21 00:00:56,250 --> 00:00:56,460 Now. 22 00:00:56,460 --> 00:01:00,310 We've got the JBOD paradigm, in other words, just a bunch of disks. 23 00:01:00,310 --> 00:01:05,240 You don't have to necessarily buy a separate direct attached storage enclosure, 24 00:01:05,240 --> 00:01:09,910 although Storage Spaces does work just fine with external storage enclosures. 25 00:01:09,910 --> 00:01:14,720 But it was built by Microsoft with the idea in mind that a 26 00:01:14,720 --> 00:01:17,850 server has N number of fixed drives, 27 00:01:17,850 --> 00:01:20,610 maybe some direct attached storage using SATA, 28 00:01:20,610 --> 00:01:23,440 SAS, USB, could be Fibre Channel, 29 00:01:23,440 --> 00:01:26,470 and we want to do more advanced volume management 30 00:01:26,470 --> 00:01:30,310 operations like tiered storage, that is combining SSD, 31 00:01:30,310 --> 00:01:33,630 solid‑state disk, and non‑SSD mechanical storage, 32 00:01:33,630 --> 00:01:36,900 and being able to have that separation of duties such 33 00:01:36,900 --> 00:01:39,930 that we can target high input/output, or I/O, 34 00:01:39,930 --> 00:01:45,840 workloads on the SSD and keep more static assets on the non‑SSD drives because 35 00:01:45,840 --> 00:01:49,340 presumably they are less expensive and so on and so forth. 36 00:01:49,340 --> 00:01:53,670 We also can do both fixed and thin volume provisioning with Storage Spaces. 37 00:01:53,670 --> 00:01:58,190 So we can create a 120‑GB virtual disk that 38 00:01:58,190 --> 00:02:01,350 actually doesn't consume that 120 GB, 39 00:02:01,350 --> 00:02:05,940 but only consumes as much spaces that you're actually using on that volume. 40 00:02:05,940 --> 00:02:09,990 Now I mentioned RAID‑like mirroring and parity because with Storage Spaces, 41 00:02:09,990 --> 00:02:14,650 those mirroring and parity and hot spare options aren't technically RAID. 42 00:02:14,650 --> 00:02:18,120 It's proprietary to the Storage Spaces product, 43 00:02:18,120 --> 00:02:23,290 but they kind of mirror, pun intended, the classic RAID levels like RAID 0, 44 00:02:23,290 --> 00:02:27,020 just a simple stripe set, RAID 1, which would be simple mirroring, 45 00:02:27,020 --> 00:02:31,320 RAID 5 that allows the failure of one disk in the array, 46 00:02:31,320 --> 00:02:35,240 and you can dynamically rebuild it after you replace the failed disk. 47 00:02:35,240 --> 00:02:39,980 It has that capability still, it's just implemented in a more proprietary way. 48 00:02:39,980 --> 00:02:44,220 Because we're talking about Windows, we have integration with Volume Shadow Copy, 49 00:02:44,220 --> 00:02:49,110 so there's no problem backing up your workloads that are on Storage Spaces. 50 00:02:49,110 --> 00:02:53,140 And we also have, very importantly, integration with failover clustering. 51 00:02:53,140 --> 00:02:57,150 Now you can use Storage Spaces itself to manage volumes in 52 00:02:57,150 --> 00:03:00,270 Windows Server independent of a failover cluster. 53 00:03:00,270 --> 00:03:04,110 In fact, if you don't know, Storage Spaces is also available on Windows client. 54 00:03:04,110 --> 00:03:07,180 But when we use Storage Spaces in a failover cluster, 55 00:03:07,180 --> 00:03:10,880 this is called Storage Spaces Direct, or S2D, 56 00:03:10,880 --> 00:03:12,870 and its value proposition, again, 57 00:03:12,870 --> 00:03:15,520 mainly it's cost savings that you don't have to have a 58 00:03:15,520 --> 00:03:18,620 dedicated external shared storage array, 59 00:03:18,620 --> 00:03:20,200 like a storage area network. 60 00:03:20,200 --> 00:03:20,790 In other words, 61 00:03:20,790 --> 00:03:23,780 you can have each node in the cluster contribute 62 00:03:23,780 --> 00:03:27,990 its own storage to a shared pool, and then using Cluster Shared Volumes, 63 00:03:27,990 --> 00:03:32,480 or CSVs, each server will have a local representation of that volume. 64 00:03:32,480 --> 00:03:34,960 This makes it much easier to pass the torch, 65 00:03:34,960 --> 00:03:38,730 as it were, when you have a clustered workload that needs a new host. 66 00:03:38,730 --> 00:03:42,360 There are three Storage Spaces resiliency types that you 67 00:03:42,360 --> 00:03:46,130 should be aware of for your career, as well as for clearing AZ‑800. 68 00:03:46,130 --> 00:03:47,010 There's mirror. 69 00:03:47,010 --> 00:03:49,070 This is where you, in a sense, 70 00:03:49,070 --> 00:03:54,080 give up 50% of the storage space on your volume because Storage 71 00:03:54,080 --> 00:03:57,760 Spaces and Windows Server will do a bit‑by‑bit copy on the 72 00:03:57,760 --> 00:03:59,720 primary virtual disk to the secondary. 73 00:03:59,720 --> 00:04:02,430 And you can do both two and three‑way mirrors. 74 00:04:02,430 --> 00:04:04,580 This is particularly important, for instance, 75 00:04:04,580 --> 00:04:07,560 for protecting your operating system disk so that if you 76 00:04:07,560 --> 00:04:11,200 do have a failure on your system drive, that is the physical drive, 77 00:04:11,200 --> 00:04:15,120 you can point to a mirror copy to keep that server up and running, 78 00:04:15,120 --> 00:04:17,350 so it's a high availability technique. 79 00:04:17,350 --> 00:04:19,440 We also have single and dual parity. 80 00:04:19,440 --> 00:04:21,670 This deals with not only striping, 81 00:04:21,670 --> 00:04:24,530 which improves I/O because you're doing reads and 82 00:04:24,530 --> 00:04:27,610 writes across two or more virtual disks, 83 00:04:27,610 --> 00:04:30,500 but then you have the ability, as I mentioned, 84 00:04:30,500 --> 00:04:35,540 to lose a physical disk and swap in a replacement and then have Storage 85 00:04:35,540 --> 00:04:40,780 Spaces dynamically rebuild the data that was on that failed disk because it's 86 00:04:40,780 --> 00:04:43,400 redundantly stored on other members in the array. 87 00:04:43,400 --> 00:04:45,040 That's the idea of parity. 88 00:04:45,040 --> 00:04:49,930 And you can survive one or even two members of that array to go down 89 00:04:49,930 --> 00:04:52,600 depending upon the parity type that you choose. 90 00:04:52,600 --> 00:04:57,780 And simple is RAID 0 striping, again, where you're combining multiple 91 00:04:57,780 --> 00:05:01,480 physical disks and you're striping data. In other words, 92 00:05:01,480 --> 00:05:05,540 your data reads and data writes are going in parallel. So hopefully you 93 00:05:05,540 --> 00:05:13,000 can see that you can dramatically increase disk I/O as opposed to doing all your reads and writes on a single disk.