1 00:00:00,940 --> 00:00:04,010 Let's go into cluster storage a little bit deeper. 2 00:00:04,010 --> 00:00:07,740 It's like peeling away layers of the proverbial onion. 3 00:00:07,740 --> 00:00:12,580 I want to double‑click into the Windows failover clustering storage options. 4 00:00:12,580 --> 00:00:17,220 I want to contrast the traditional shared storage model with 5 00:00:17,220 --> 00:00:21,540 Microsoft's proprietary Storage Spaces Direct technology. 6 00:00:21,540 --> 00:00:23,700 When I'm talking traditional shared storage, 7 00:00:23,700 --> 00:00:28,040 I'm talking about where you have dedicated hardware arrays of storage, 8 00:00:28,040 --> 00:00:32,540 a storage area network, block level network access storage. 9 00:00:32,540 --> 00:00:36,240 The disks are normally serial attached SCSI, 10 00:00:36,240 --> 00:00:42,040 and the connectivity is normally either Fibre Channel or iSCSI. 11 00:00:42,040 --> 00:00:46,040 Now this is going to give you speed, scalability, security. 12 00:00:46,040 --> 00:00:49,960 The storage array may have their own backup and recovery options, 13 00:00:49,960 --> 00:00:52,460 they may have their own resiliency options, 14 00:00:52,460 --> 00:00:56,210 but you're going to pay because it's extra money for that storage. 15 00:00:56,210 --> 00:00:58,580 It's learning curve, it's job role stuff, 16 00:00:58,580 --> 00:01:01,840 so there is a lot of overhead with this option. 17 00:01:01,840 --> 00:01:06,080 Multipath I/O is a Windows Server feature where you could have 18 00:01:06,080 --> 00:01:09,530 your SAN with multiple network interconnects, 19 00:01:09,530 --> 00:01:13,540 redundant interconnects to where if one connection between a server 20 00:01:13,540 --> 00:01:17,650 across the network to a storage array LUN were to fail, 21 00:01:17,650 --> 00:01:20,740 the server could use a backup path. 22 00:01:20,740 --> 00:01:23,880 Another software feature of Windows Server is the 23 00:01:23,880 --> 00:01:26,770 idea of network adapter teaming, same idea. 24 00:01:26,770 --> 00:01:29,620 So these are just Windows Server features that you can bring to 25 00:01:29,620 --> 00:01:32,740 bear to increase the availability and speed. 26 00:01:32,740 --> 00:01:35,670 So that works fine with Windows failover clustering, 27 00:01:35,670 --> 00:01:38,640 but you're going to, on your exam, need to know about, 28 00:01:38,640 --> 00:01:42,120 more about, much more about Storage Spaces Direct, 29 00:01:42,120 --> 00:01:45,130 and some value props here, as I alluded to earlier, 30 00:01:45,130 --> 00:01:48,740 you can use serial‑attached SCSI, just a bunch of disks, 31 00:01:48,740 --> 00:01:50,750 or JBOD enclosures, in other words, 32 00:01:50,750 --> 00:01:53,810 we're talking about commodity hardware where we can 33 00:01:53,810 --> 00:01:57,270 combine that if your cluster nodes have fairly recent 34 00:01:57,270 --> 00:02:01,450 network interfaces that support SMB, Server Message Block 3, 35 00:02:01,450 --> 00:02:04,940 and Remote Direct Memory Access, or RDMA. 36 00:02:04,940 --> 00:02:09,540 You can overcome the difference in speed that you might normally expect 37 00:02:09,540 --> 00:02:13,690 between servers using their own storage and their own direct attached 38 00:02:13,690 --> 00:02:17,270 storage and a high speed SAN as an alternative. 39 00:02:17,270 --> 00:02:20,250 I mean, let's face it, if you were to do a benchmark, 40 00:02:20,250 --> 00:02:24,540 your traditional shared storage is probably going to win every day, 41 00:02:24,540 --> 00:02:28,920 but there are benefits to Storage Spaces Direct beyond just the fact 42 00:02:28,920 --> 00:02:31,740 you don't have to have all that extra infrastructure. 43 00:02:31,740 --> 00:02:35,090 You can layer in some of these native Windows Server technologies. 44 00:02:35,090 --> 00:02:38,710 Cluster shared volumes is a nice efficient way to 45 00:02:38,710 --> 00:02:41,690 surface your shared storage that, again, 46 00:02:41,690 --> 00:02:45,040 is directly attached to each cluster node, 47 00:02:45,040 --> 00:02:49,030 and you can surface those logical volumes on all nodes to 48 00:02:49,030 --> 00:02:51,940 make that roll switch over even faster. 49 00:02:51,940 --> 00:02:54,940 Again, multipath I/O, teamed NICs. 50 00:02:54,940 --> 00:02:57,980 So I just want you to go back and forth kind of A/B test. 51 00:02:57,980 --> 00:03:02,440 If your business already has SAN, well your decision may be made, 52 00:03:02,440 --> 00:03:05,210 but for the exam, I just can't emphasize this enough. 53 00:03:05,210 --> 00:03:08,230 You really need to be on board with Storage Spaces Direct, 54 00:03:08,230 --> 00:03:17,000 and I want to encourage you that you can set up S2D and failover clustering all on your own lab environment using Hyper‑V.