1 00:00:06,720 --> 00:00:08,700 - [Instructor] There are many types of RAID 2 00:00:08,700 --> 00:00:12,363 as shown by the myriad versions available, 3 00:00:13,350 --> 00:00:18,350 RAID zero, RAID one, five, six, one plus zero, 4 00:00:19,290 --> 00:00:21,240 and on and on and on. 5 00:00:21,240 --> 00:00:24,300 You rarely see two through four. 6 00:00:24,300 --> 00:00:27,990 Those are parity types of RAID that have been completely 7 00:00:27,990 --> 00:00:31,743 eclipsed by five and six so you rarely see those. 8 00:00:32,700 --> 00:00:35,550 And while they are all redundant, 9 00:00:35,550 --> 00:00:39,330 not all are truly fault tolerant. 10 00:00:39,330 --> 00:00:42,270 Redundant means that you have more than one. 11 00:00:42,270 --> 00:00:47,160 Fault tolerant means that you can survive a failure 12 00:00:47,160 --> 00:00:49,260 and still get access to your data. 13 00:00:49,260 --> 00:00:51,060 And that's what you want. 14 00:00:51,060 --> 00:00:55,050 For example, RAID zero is not fault tolerant. 15 00:00:55,050 --> 00:00:57,390 RAID zero is just known as striping. 16 00:00:57,390 --> 00:01:00,390 You Stripe data across multiple drives 17 00:01:00,390 --> 00:01:03,300 to increase speed and performance. 18 00:01:03,300 --> 00:01:05,040 So we don't really use RAID zero 19 00:01:05,040 --> 00:01:07,260 if we want something fault tolerant, 20 00:01:07,260 --> 00:01:09,840 because if one drive was to fail, 21 00:01:09,840 --> 00:01:11,820 we would lose all of the data. 22 00:01:11,820 --> 00:01:15,240 Instead, we'll go for RAID types such as RAID one, 23 00:01:15,240 --> 00:01:18,060 which is mirroring and RAID five which 24 00:01:18,060 --> 00:01:20,130 is striping with parity. 25 00:01:20,130 --> 00:01:24,120 These are very common types of fault tolerant RAID. 26 00:01:24,120 --> 00:01:28,410 They can survive a fault or a failure 27 00:01:28,410 --> 00:01:31,473 and keep supplying data to the users. 28 00:01:32,460 --> 00:01:35,463 Let's describe those two in a little detail now. 29 00:01:37,380 --> 00:01:41,010 RAID one is known as mirroring. 30 00:01:41,010 --> 00:01:43,350 It requires two drives which should be 31 00:01:43,350 --> 00:01:45,810 of the same make and model. 32 00:01:45,810 --> 00:01:50,250 For example, you would want the same Western digital drives, 33 00:01:50,250 --> 00:01:52,563 within the same blue series. 34 00:01:53,670 --> 00:01:57,420 All data is written to both of the drives. 35 00:01:57,420 --> 00:01:58,563 It's mirrored. 36 00:01:59,580 --> 00:02:02,220 If one drive fails, then the other continues 37 00:02:02,220 --> 00:02:05,583 to function offering that fault tolerance that you want. 38 00:02:07,260 --> 00:02:11,250 Data is normally read from only one drive. 39 00:02:11,250 --> 00:02:14,190 So in that respect, you get the same performance 40 00:02:14,190 --> 00:02:17,130 as you would in a single drive system, 41 00:02:17,130 --> 00:02:19,710 but when you write the data to the mirror, 42 00:02:19,710 --> 00:02:22,080 it takes a little longer 'cause you're writing 43 00:02:22,080 --> 00:02:24,243 to two drives at the same time. 44 00:02:25,380 --> 00:02:27,690 Let's show an illustration of this. 45 00:02:27,690 --> 00:02:31,290 Here, we have two discs, disc one and disc two each 46 00:02:31,290 --> 00:02:34,470 of which are 500 gigabytes and they're connected 47 00:02:34,470 --> 00:02:37,170 to a RAID controller. 48 00:02:37,170 --> 00:02:41,130 And quite often the controller is just an adapter card, 49 00:02:41,130 --> 00:02:44,460 could be a PCI express by one card 50 00:02:44,460 --> 00:02:48,570 or it could be an integrated device within your motherboard. 51 00:02:48,570 --> 00:02:52,230 When the mirror is created, the two drives 52 00:02:52,230 --> 00:02:54,060 basically become one. 53 00:02:54,060 --> 00:02:58,050 They're collectively known as a single volume letter. 54 00:02:58,050 --> 00:03:01,020 In this case, we chose M, M for mirror, 55 00:03:01,020 --> 00:03:02,670 but you can choose whatever letter you want 56 00:03:02,670 --> 00:03:04,980 as long as it's not been used yet. 57 00:03:04,980 --> 00:03:08,430 One of the issues you might find with RAID one arrays, 58 00:03:08,430 --> 00:03:12,210 is that the controller acts as a single point of failure. 59 00:03:12,210 --> 00:03:17,210 And so you could create a new configuration where each disc 60 00:03:17,340 --> 00:03:20,760 is connected to a separate controller. 61 00:03:20,760 --> 00:03:23,940 And this is known as disc duplexing. 62 00:03:23,940 --> 00:03:26,860 So you have even more fault tolerance at this point 63 00:03:28,650 --> 00:03:32,010 RAID arrays can be controlled through hardware 64 00:03:32,010 --> 00:03:33,453 or through software. 65 00:03:34,500 --> 00:03:38,730 An example of hardware would be using a RAID adapter card 66 00:03:38,730 --> 00:03:42,900 as we mentioned before, where a RAID enabled motherboard 67 00:03:42,900 --> 00:03:45,600 and the array is created within the card 68 00:03:45,600 --> 00:03:47,610 or the motherboard's firmware. 69 00:03:47,610 --> 00:03:50,913 This is done outside of the operating system. 70 00:03:52,020 --> 00:03:54,510 An example of software would be to create 71 00:03:54,510 --> 00:03:57,600 the array in compliant versions of Windows, 72 00:03:57,600 --> 00:04:00,693 say within the disc management console Window. 73 00:04:02,520 --> 00:04:05,550 Let's show an actual RAID one array 74 00:04:05,550 --> 00:04:07,380 that was created within Windows 75 00:04:07,380 --> 00:04:08,880 and then we'll show one that's created 76 00:04:08,880 --> 00:04:11,163 in firmware on a separate system. 77 00:04:14,400 --> 00:04:18,240 Here's an example of a RAID one array controlled 78 00:04:18,240 --> 00:04:20,670 by software in Windows. 79 00:04:20,670 --> 00:04:22,380 And it's controlled within the disc 80 00:04:22,380 --> 00:04:24,540 management console Window. 81 00:04:24,540 --> 00:04:27,330 Here, we have our list of volumes, for example, 82 00:04:27,330 --> 00:04:30,930 the C volume which is on its own hard drive. 83 00:04:30,930 --> 00:04:33,303 And that's where the operating system lives. 84 00:04:34,170 --> 00:04:37,380 And then we have the M volume which is the mirror 85 00:04:37,380 --> 00:04:41,010 and that is comprised of two other discs, 86 00:04:41,010 --> 00:04:43,260 disc two and disc three. 87 00:04:43,260 --> 00:04:46,230 And you can see here, the reddish brown header, 88 00:04:46,230 --> 00:04:48,960 indicates a mirrored volume. 89 00:04:48,960 --> 00:04:51,090 So M is the mirrored volume. 90 00:04:51,090 --> 00:04:54,630 Any data that is written to this volume is written 91 00:04:54,630 --> 00:04:57,303 to both drives simultaneously. 92 00:04:58,530 --> 00:05:00,120 When it's read from the volume, 93 00:05:00,120 --> 00:05:03,840 it just reads from the first volume in the mirror, disc two. 94 00:05:03,840 --> 00:05:06,450 If one disc was to fail and most likely 95 00:05:06,450 --> 00:05:09,300 that's the lower number in the mirror, 96 00:05:09,300 --> 00:05:11,610 disc two would probably be the first to fail, 97 00:05:11,610 --> 00:05:14,970 but if that was to fail, then the mirror would be broken. 98 00:05:14,970 --> 00:05:18,450 However, users would still be able to read 99 00:05:18,450 --> 00:05:21,120 and write data from disc three. 100 00:05:21,120 --> 00:05:23,620 The M volume would continue to function 101 00:05:24,600 --> 00:05:28,410 and that is the fault tolerant portion of RAID one, 102 00:05:28,410 --> 00:05:29,670 but we would need to fix it. 103 00:05:29,670 --> 00:05:34,470 We'd need to replace that disc two physically 104 00:05:34,470 --> 00:05:38,370 and then rebuild the mirror which would then copy all 105 00:05:38,370 --> 00:05:42,750 the data from disc three to disc two 106 00:05:42,750 --> 00:05:44,790 and then it would resynchronize 107 00:05:44,790 --> 00:05:46,740 and then we would have our mirror back. 108 00:05:48,300 --> 00:05:51,180 Now, here's an example of a RAID one array 109 00:05:51,180 --> 00:05:53,550 controlled by firmware. 110 00:05:53,550 --> 00:05:56,640 This is Intel Rapid Storage Technology. 111 00:05:56,640 --> 00:06:01,640 It's a built in RAID controller within the motherboard. 112 00:06:01,860 --> 00:06:05,240 This motherboard can handle RAID one and can do RAID five. 113 00:06:05,240 --> 00:06:08,340 In this case, I have a RAID one array. 114 00:06:08,340 --> 00:06:13,340 There's two Western digital drives working in concert. 115 00:06:13,680 --> 00:06:16,860 Each one is one terabyte that shows up here 116 00:06:16,860 --> 00:06:19,290 as 931 gigabytes. 117 00:06:19,290 --> 00:06:22,563 And it shows you that it is indeed a RAID one mirror. 118 00:06:23,460 --> 00:06:28,460 If this mirror was to break, if it was to fail, 119 00:06:29,010 --> 00:06:31,320 we would have to fix it here. 120 00:06:31,320 --> 00:06:34,890 We would have to do recovery here and not in Windows. 121 00:06:34,890 --> 00:06:38,520 If we were to take a look at the volume that was created 122 00:06:38,520 --> 00:06:43,410 for this RAID one mirror in Windows in disc management, 123 00:06:43,410 --> 00:06:46,770 we would see it would just be one drive. 124 00:06:46,770 --> 00:06:50,340 It would show up as one disc, not as two 125 00:06:50,340 --> 00:06:51,597 as you can see here. 126 00:06:51,597 --> 00:06:54,480 And this is the disc management Window again. 127 00:06:54,480 --> 00:06:59,480 Data E is that mirror, but it only shows as one disc. 128 00:07:00,720 --> 00:07:03,330 So you don't actually know that you have a 129 00:07:03,330 --> 00:07:05,100 mirror running here. 130 00:07:05,100 --> 00:07:07,560 If you look at it from Windows, 131 00:07:07,560 --> 00:07:11,100 you would have to look at the discs individually 132 00:07:11,100 --> 00:07:14,010 within the BIOS or within the firmware. 133 00:07:14,010 --> 00:07:16,290 And again, if it was to fail, you'd have 134 00:07:16,290 --> 00:07:19,323 to rebuild it within the firmware itself. 135 00:07:20,190 --> 00:07:22,950 And to get into that firmware for Intel, 136 00:07:22,950 --> 00:07:25,560 it's usually Control + I, and you press 137 00:07:25,560 --> 00:07:28,500 that before you would even get into the BIOS. 138 00:07:28,500 --> 00:07:31,050 Your F2 to get into an Intel BIOS. 139 00:07:31,050 --> 00:07:35,640 You press the Control + I to get into the firmware 140 00:07:35,640 --> 00:07:39,030 for the RAID adapter before that. 141 00:07:39,030 --> 00:07:41,220 And that's again, Control + I. 142 00:07:41,220 --> 00:07:42,813 So that's RAID one. 143 00:07:44,220 --> 00:07:47,010 Now, we move into RAID five, which is known 144 00:07:47,010 --> 00:07:49,860 as striping with parity. 145 00:07:49,860 --> 00:07:54,540 With RAID zero, you Stripe data across multiple drives 146 00:07:54,540 --> 00:07:56,310 to increase performance. 147 00:07:56,310 --> 00:07:58,860 And you do this with RAID five as well, 148 00:07:58,860 --> 00:08:03,450 but now we also use parity so that in the event 149 00:08:03,450 --> 00:08:06,870 of data failure we can use the parity information 150 00:08:06,870 --> 00:08:08,883 to recreate that data. 151 00:08:11,100 --> 00:08:14,010 It requires a minimum of three drives, 152 00:08:14,010 --> 00:08:16,590 which should be of the same make and model. 153 00:08:16,590 --> 00:08:21,060 RAID five arrays often have three, four or five drives, 154 00:08:21,060 --> 00:08:22,650 but you're not limited to that. 155 00:08:22,650 --> 00:08:24,660 I believe the normal maximum for hardware 156 00:08:24,660 --> 00:08:26,943 and software is 32 drives. 157 00:08:28,650 --> 00:08:32,670 Data is striped across all drives regardless 158 00:08:32,670 --> 00:08:34,770 of how many drives you have. 159 00:08:34,770 --> 00:08:38,553 Again, it's done in an effort to increase performance. 160 00:08:40,140 --> 00:08:43,950 Parity information is distributed across drives 161 00:08:43,950 --> 00:08:46,680 in a staggered format. 162 00:08:46,680 --> 00:08:50,580 This is known as distributed parity blocks. 163 00:08:50,580 --> 00:08:54,570 Any block of data also has a calculated amount 164 00:08:54,570 --> 00:08:57,960 of parity and that's for repairing the data 165 00:08:57,960 --> 00:09:00,543 if that particular drive fails. 166 00:09:01,560 --> 00:09:04,200 This parity information is distributed among all 167 00:09:04,200 --> 00:09:07,200 the other drives, except for the one where 168 00:09:07,200 --> 00:09:09,303 the data was written to. 169 00:09:11,730 --> 00:09:15,690 So in this scenario, the data is divided into three parts, 170 00:09:15,690 --> 00:09:17,850 A, B and C. 171 00:09:17,850 --> 00:09:22,200 Let's imagine that we have a 384 kilobyte file 172 00:09:22,200 --> 00:09:24,603 that we want to write to the array. 173 00:09:25,530 --> 00:09:27,870 Well, that's going to be written in blocks 174 00:09:27,870 --> 00:09:30,810 to each drive one at a time. 175 00:09:30,810 --> 00:09:34,413 And each of these blocks would be 64 kilobytes. 176 00:09:36,330 --> 00:09:41,110 So for example, drive one and drive two will be written to 177 00:09:41,970 --> 00:09:46,650 and A1 is a 64 kilobyte block. 178 00:09:46,650 --> 00:09:50,820 A2 is another 64 kilobyte block. 179 00:09:50,820 --> 00:09:54,450 And these are also sometimes referred to as chunks. 180 00:09:54,450 --> 00:09:57,273 Together, they are a Stripe of data. 181 00:09:58,920 --> 00:10:01,260 And it continues in that manner. 182 00:10:01,260 --> 00:10:06,000 The next 64 kilobyte block would be B1 and B2, 183 00:10:06,000 --> 00:10:09,063 but this time they're written to drive one and drive three. 184 00:10:10,500 --> 00:10:14,580 And then the C block of data, C1 and C2 which are written 185 00:10:14,580 --> 00:10:16,983 to drive two and drive three. 186 00:10:17,970 --> 00:10:21,120 In total, we have six blocks of data. 187 00:10:21,120 --> 00:10:24,360 And if you multiply that by 64 kilobytes each, 188 00:10:24,360 --> 00:10:28,680 that gives us our total of 384 kilobytes, 189 00:10:28,680 --> 00:10:31,050 which is the size of the file. 190 00:10:31,050 --> 00:10:33,270 The parity for each Stripe of data 191 00:10:33,270 --> 00:10:36,480 is placed on the remaining drive. 192 00:10:36,480 --> 00:10:39,900 Now, in an effort to protect the data in the event 193 00:10:39,900 --> 00:10:42,750 of failure, we're using this parity. 194 00:10:42,750 --> 00:10:46,230 So parity for each Stripe of data is placed 195 00:10:46,230 --> 00:10:49,890 on that remaining drive for each letter, 196 00:10:49,890 --> 00:10:52,890 for each set of blocks, basically. 197 00:10:52,890 --> 00:10:55,350 And you can think of parity information as a sort 198 00:10:55,350 --> 00:10:59,400 of compressed version of the Stripe of data. 199 00:10:59,400 --> 00:11:04,400 So the A1 and A2 Stripe of data has a parity version 200 00:11:04,710 --> 00:11:08,133 on drive three which we will call AP. 201 00:11:09,150 --> 00:11:14,150 The B1 and B2 Stripe of data has a parity version 202 00:11:14,400 --> 00:11:17,643 on drive two, we'll call that BP. 203 00:11:18,840 --> 00:11:23,840 And the C1 and C2 Stripe of data has parity on drive one 204 00:11:24,240 --> 00:11:26,610 and we'll call that CP. 205 00:11:26,610 --> 00:11:29,820 You'll notice that the parity blocks are staggered. 206 00:11:29,820 --> 00:11:34,820 It starts with drive three, then drive two, then drive one. 207 00:11:35,010 --> 00:11:38,793 You don't want to have all of the parity on one drive. 208 00:11:39,630 --> 00:11:44,130 So this way we can recreate any data block or portion 209 00:11:44,130 --> 00:11:49,130 of that data block if it fails or if an entire drive fails. 210 00:11:50,280 --> 00:11:54,120 So again, in a RAID five environment, 211 00:11:54,120 --> 00:11:57,810 one hard drive can fail and you'll still be able 212 00:11:57,810 --> 00:11:59,640 to get access to the data. 213 00:11:59,640 --> 00:12:02,250 And we could still recreate the data once we replace 214 00:12:02,250 --> 00:12:05,550 that drive, but if a second drive was to fail, 215 00:12:05,550 --> 00:12:06,900 you'd be in trouble. 216 00:12:06,900 --> 00:12:09,090 That's why there's RAID six. 217 00:12:09,090 --> 00:12:11,970 With RAID six, you have additional parity 218 00:12:11,970 --> 00:12:15,870 and up to two drives can fail while still 219 00:12:15,870 --> 00:12:17,610 being fault tolerant. 220 00:12:17,610 --> 00:12:20,430 So you have to decide how much fault tolerance do we need? 221 00:12:20,430 --> 00:12:22,560 How much money do I have for hard drives? 222 00:12:22,560 --> 00:12:23,970 And that will help you decide whether 223 00:12:23,970 --> 00:12:26,403 you're gonna use RAID five or RAID six. 224 00:12:27,630 --> 00:12:30,150 Here's an example of a RAID five array that 225 00:12:30,150 --> 00:12:31,620 is controlled by software. 226 00:12:31,620 --> 00:12:35,160 It's running in Windows in the disc management Window 227 00:12:35,160 --> 00:12:36,870 and we have our three drives here. 228 00:12:36,870 --> 00:12:39,930 They're labeled as disc one, disc two and disc three. 229 00:12:39,930 --> 00:12:43,440 Collectively, they make up the S volume 230 00:12:43,440 --> 00:12:46,890 and that light blue, that baby blue 231 00:12:46,890 --> 00:12:48,900 in the header means RAID five volume 232 00:12:48,900 --> 00:12:51,060 for this operating system. 233 00:12:51,060 --> 00:12:54,060 If you write data to this volume, 234 00:12:54,060 --> 00:12:57,900 it is striped across the three drives, 235 00:12:57,900 --> 00:13:00,960 but parity information is included as well 236 00:13:00,960 --> 00:13:02,820 for each data block. 237 00:13:02,820 --> 00:13:04,920 If one drive was to fail, you would need 238 00:13:04,920 --> 00:13:07,590 to replace that drive and then rebuild 239 00:13:07,590 --> 00:13:10,020 the array within disc management 240 00:13:10,020 --> 00:13:13,980 and rebuild the data on new drive, 241 00:13:13,980 --> 00:13:17,070 or you could attempt to restore from backup. 242 00:13:17,070 --> 00:13:21,510 Now, software's great, but hardware is always superior. 243 00:13:21,510 --> 00:13:23,610 And you're definitely gonna see hardware based 244 00:13:23,610 --> 00:13:25,440 RAID five arrays. 245 00:13:25,440 --> 00:13:27,600 And here's an example of some hardware 246 00:13:27,600 --> 00:13:30,450 that you might see used in the array. 247 00:13:30,450 --> 00:13:33,570 There is a controller at the top which can handle 248 00:13:33,570 --> 00:13:38,280 as many as five drives and it's an SATA controller. 249 00:13:38,280 --> 00:13:41,430 And on the bottom, we have three hard drives 250 00:13:41,430 --> 00:13:43,533 that will be connected to the controller. 251 00:13:44,760 --> 00:13:47,460 And again, you have to have three drives minimum 252 00:13:47,460 --> 00:13:49,770 and they should be the same make and model. 253 00:13:49,770 --> 00:13:54,120 The controllers firmware could be accessed previous 254 00:13:54,120 --> 00:13:57,060 to the operating system being booted in the same manner 255 00:13:57,060 --> 00:14:02,060 that we pressed Control + I to get into the RAID firmware 256 00:14:02,400 --> 00:14:04,200 on the Intel motherboard. 257 00:14:04,200 --> 00:14:06,180 But again, you wanna make sure the make and model 258 00:14:06,180 --> 00:14:08,940 of the hard drives is the same. 259 00:14:08,940 --> 00:14:12,630 And if you look closely at these two drives, 260 00:14:12,630 --> 00:14:15,150 you'll see a slight difference. 261 00:14:15,150 --> 00:14:20,150 One, is a 1600AVBS, and the other is a 1600AVJS. 262 00:14:23,490 --> 00:14:27,480 They're both 160 gig drives, but they're slightly different. 263 00:14:27,480 --> 00:14:29,490 And I think the one difference with these 264 00:14:29,490 --> 00:14:32,310 is the cache memory on board. 265 00:14:32,310 --> 00:14:33,870 And that could be a big deal. 266 00:14:33,870 --> 00:14:36,540 You could try it, you could see if it'll work, 267 00:14:36,540 --> 00:14:38,880 but it's not really a good idea. 268 00:14:38,880 --> 00:14:41,130 In addition to that, you see the drive on the left 269 00:14:41,130 --> 00:14:44,910 has been written on, that's always a bad sign as well. 270 00:14:44,910 --> 00:14:47,250 So when you're working with hard drives, 271 00:14:47,250 --> 00:14:51,630 work with new drives, don't just grab anything you can find 272 00:14:51,630 --> 00:14:52,740 that's already been used. 273 00:14:52,740 --> 00:14:54,990 You don't know the condition of that drive. 274 00:14:54,990 --> 00:14:57,540 You wanna use new drives and know 275 00:14:57,540 --> 00:14:59,850 that you're gonna have a shelf life of say two 276 00:14:59,850 --> 00:15:02,490 or three years for those drives. 277 00:15:02,490 --> 00:15:04,890 And you want to know that they're all new, 278 00:15:04,890 --> 00:15:08,730 newly formatted and brand new outta the box. 279 00:15:08,730 --> 00:15:11,220 So watch out for any drives that have been written on, 280 00:15:11,220 --> 00:15:12,810 watch out for slight changes in 281 00:15:12,810 --> 00:15:15,450 the name, slight differences. 282 00:15:15,450 --> 00:15:16,950 Make sure you're using the same drive's 283 00:15:16,950 --> 00:15:18,870 manufacturer and model. 284 00:15:18,870 --> 00:15:21,070 And that's about the end of this sub lesson.