1 00:00:00,050 --> 00:00:03,260 On this episode of restore it all we talk about what I think is the 2 00:00:03,260 --> 00:00:06,830 biggest advancement in backup and recovery technology during my career. 3 00:00:07,100 --> 00:00:08,510 And that's deduplication. 4 00:00:08,720 --> 00:00:09,980 I hope you enjoy the episode. 5 00:00:29,768 --> 00:00:32,468 W. Curtis Preston: Hi, and welcome to Backup Central's Restored all podcast. 6 00:00:33,148 --> 00:00:35,698 I'm your host, w Curtis Preston, aka Mr. 7 00:00:35,698 --> 00:00:35,848 Backup. 8 00:00:36,703 --> 00:00:42,913 And a half with me, my network, rearchitect Rearchitect, engineer. 9 00:00:45,113 --> 00:00:47,773 Prasanna Malaiyandi: Hey, Curtis, whatever I could do to keep you safe, you know? 10 00:00:49,363 --> 00:00:53,053 W. Curtis Preston: You know what's really funny is like I, I consider myself a 11 00:00:53,058 --> 00:01:01,003 pretty tech savvy guy, and when we were talking today, About what I'm, you know 12 00:01:01,003 --> 00:01:04,963 how I've, I've replaced a bunch of gear and I'm swapping out some stuff and 13 00:01:04,963 --> 00:01:08,803 moving some cables around, and then you were like, you were yelling at me. 14 00:01:08,803 --> 00:01:10,873 You were like, you can't do that. 15 00:01:11,173 --> 00:01:14,023 You can't put the switch on the thing. 16 00:01:14,023 --> 00:01:16,633 And I was like, yeah, I can, like, what are you talking about? 17 00:01:16,633 --> 00:01:21,433 And it, and it took me like a couple of seconds and I was like, oh, wait. 18 00:01:21,433 --> 00:01:21,973 You're right. 19 00:01:22,033 --> 00:01:24,883 I can't, that's not, I can't do that. 20 00:01:24,883 --> 00:01:25,663 I can't put. 21 00:01:27,328 --> 00:01:28,138 The switch. 22 00:01:28,198 --> 00:01:30,328 I can't put the router. 23 00:01:31,018 --> 00:01:33,748 That's gonna be my firewall on the same switch 24 00:01:34,198 --> 00:01:35,758 As my home LAN 25 00:01:35,863 --> 00:01:36,253 Prasanna Malaiyandi: Yeah, 26 00:01:37,348 --> 00:01:39,918 W. Curtis Preston: I dunno what I was thinking. 27 00:01:42,058 --> 00:01:42,238 Yeah. 28 00:01:42,238 --> 00:01:45,873 Prasanna Malaiyandi: Just another topic that I know just a little bit about. 29 00:01:48,958 --> 00:01:51,838 W. Curtis Preston: I'm a little, I feel a little ashamed that that was. 30 00:01:53,173 --> 00:01:56,953 But I'm glad I talked to you about my, you know, as, as is 31 00:01:56,953 --> 00:01:58,573 the case with many subjects. 32 00:01:59,053 --> 00:02:02,578 I'm glad I talked to you about, you know, what I'm up to. 33 00:02:02,718 --> 00:02:03,228 Um, 34 00:02:03,908 --> 00:02:04,578 Prasanna Malaiyandi: Glad I could help. 35 00:02:05,263 --> 00:02:10,033 W. Curtis Preston: I have successfully purchased and configured for the 36 00:02:10,033 --> 00:02:12,253 video for the video watchers. 37 00:02:12,853 --> 00:02:15,373 Let's see if it makes it into the camera before the cable runs. 38 00:02:15,943 --> 00:02:23,083 There it is, the ASUS AX6600, which is a mesh router. 39 00:02:23,653 --> 00:02:27,633 And I gotta say it's much more better than what I had before, 40 00:02:28,153 --> 00:02:30,388 and it's able, I've got two. 41 00:02:30,388 --> 00:02:34,498 It's supposed to provide 5,500 square feet, but of course that's, that doesn't 42 00:02:34,498 --> 00:02:37,588 include drywall and two by fours, right? 43 00:02:37,693 --> 00:02:42,163 Prasanna Malaiyandi: it's crazy how much signal degrades going through drywall. 44 00:02:42,703 --> 00:02:45,643 And the other thing people don't realize is five gigahertz, 45 00:02:46,033 --> 00:02:48,193 like degrades like no tomorrow 46 00:02:48,688 --> 00:02:49,318 W. Curtis Preston: Right. 47 00:02:49,558 --> 00:02:53,428 Remind me, remind me why five gertz is better again. 48 00:02:54,598 --> 00:02:58,888 Prasanna Malaiyandi: It's faster because it can handle more bandwidth, and also 49 00:02:58,888 --> 00:03:03,118 the channel is wider, so you can have more things talking at the same time. 50 00:03:03,118 --> 00:03:06,208 It's just as your frequency goes up, the distance goes 51 00:03:06,208 --> 00:03:08,218 down for the same power levels, 52 00:03:08,578 --> 00:03:10,558 W. Curtis Preston: So is this like DC versus ac? 53 00:03:11,128 --> 00:03:12,688 Prasanna Malaiyandi: not quite DC versus ac. 54 00:03:13,018 --> 00:03:13,538 It's more about. 55 00:03:14,683 --> 00:03:18,553 You need to pump as many things as possible into, because high frequency, 56 00:03:18,553 --> 00:03:24,913 right, it's more per cycle, right, than 2.4, which is less airtime, if you will. 57 00:03:25,693 --> 00:03:26,173 W. Curtis Preston: Right. 58 00:03:26,593 --> 00:03:29,203 Prasanna Malaiyandi: And so every sort of peak, you can send 59 00:03:29,203 --> 00:03:32,743 more out with the five gigahertz because you're doing it more often. 60 00:03:33,613 --> 00:03:34,063 W. Curtis Preston: right. 61 00:03:35,653 --> 00:03:37,993 Prasanna Malaiyandi: And so it works a lot better. 62 00:03:37,993 --> 00:03:40,333 It's just the distance isn't as great. 63 00:03:40,693 --> 00:03:43,303 Now, I will tell people, so this is one of my, I'm gonna 64 00:03:43,308 --> 00:03:45,103 get up on my soapbox now, right? 65 00:03:45,253 --> 00:03:49,093 One of my rare soapbox events and tell people, a lot of times people 66 00:03:49,093 --> 00:03:54,133 think they need more wifi access points in their house to get coverage. 67 00:03:54,513 --> 00:03:57,853 And to those people, I will say, plan out your network carefully. 68 00:03:57,853 --> 00:03:59,593 Put your devices where they matter. 69 00:04:00,448 --> 00:04:05,338 And also don't put too many devices and don't crank up the power all the way 70 00:04:05,343 --> 00:04:08,398 to high, because I know Curtis, you and I were talking about this when you're 71 00:04:08,398 --> 00:04:12,238 looking at mesh, and it was like, imagine that your router can overpower your 72 00:04:12,238 --> 00:04:17,968 phone, your laptop, your iPad, so it's screaming at the top of its lungs and your 73 00:04:17,973 --> 00:04:20,398 phone can barely even scream back at it. 74 00:04:20,398 --> 00:04:24,448 And so that's actually worse for your network and for airtime than 75 00:04:24,448 --> 00:04:26,248 actually sort of balancing out power. 76 00:04:26,908 --> 00:04:28,918 W. Curtis Preston: I just don't know if, like, the stuff 77 00:04:28,923 --> 00:04:30,538 you're talking about, like is. 78 00:04:31,983 --> 00:04:36,113 is that even, is that configuration option even on consumer class routers? 79 00:04:36,506 --> 00:04:38,723 Prasanna Malaiyandi: you'll have sort of the low, medium, high power 80 00:04:38,728 --> 00:04:43,313 levels, uh, but it takes time to fine tune and tweet these, right? 81 00:04:43,318 --> 00:04:47,663 You have to walk around with a wifi analyzer on your phone, right? 82 00:04:47,663 --> 00:04:50,783 So Apple with their, uh, iPhones, right? 83 00:04:50,783 --> 00:04:52,193 They ship, what is it? 84 00:04:52,193 --> 00:04:54,343 Airport utility, which has a wifi scan. 85 00:04:55,033 --> 00:04:57,733 Option, which will show you all the wifi networks and sort of the signal 86 00:04:57,733 --> 00:05:00,193 strength, and you basically have to walk around your house with that and 87 00:05:00,193 --> 00:05:01,333 be like, okay, where is it strong? 88 00:05:01,333 --> 00:05:02,023 Where is it weak? 89 00:05:02,863 --> 00:05:04,183 Right, to figure out the placement. 90 00:05:04,213 --> 00:05:07,603 That's the ideal way, because what you want is you want coverage in the right 91 00:05:07,603 --> 00:05:11,743 places, because what you see is in a lot of high density housing areas, or 92 00:05:11,923 --> 00:05:16,333 even homes next to each other is most people end up with crummy wifi because 93 00:05:16,363 --> 00:05:20,953 their power is turned up so high, it bleeds into everyone else's area 94 00:05:21,073 --> 00:05:22,723 such that everyone has a crappy time. 95 00:05:23,353 --> 00:05:26,468 because then you get interference and then everyone sort of slows down and then it 96 00:05:26,528 --> 00:05:26,888 W. Curtis Preston: Right. 97 00:05:27,218 --> 00:05:27,428 Yeah. 98 00:05:27,428 --> 00:05:28,628 I got a lot of wifi. 99 00:05:28,898 --> 00:05:30,128 I got a lot of networks. 100 00:05:30,848 --> 00:05:34,898 Um, you know, um, yeah, 101 00:05:35,558 --> 00:05:38,408 Prasanna Malaiyandi: And for, and for the last bit, last bit of my 102 00:05:38,408 --> 00:05:43,268 soapbox is please, please, please do not use 40 megahertz channel widths 103 00:05:43,273 --> 00:05:45,278 on your 2.4 gigahertz channels. 104 00:05:45,518 --> 00:05:50,258 You do not need to use 40 megahertz and ruin everyone else's connectivity. 105 00:05:50,678 --> 00:05:54,248 Please only use 20 megahertz bands for 2.4 gigahertz. 106 00:05:55,703 --> 00:06:00,713 W. Curtis Preston: Uh, I'll see what I can do but I, but I have this 107 00:06:00,713 --> 00:06:05,273 new, you know, and again, I am not a wireless, I feel like a wireless nbe, 108 00:06:05,543 --> 00:06:10,733 but I have this new fancy right where it automatically selects the right. 109 00:06:10,873 --> 00:06:11,813 Um, that's 110 00:06:11,813 --> 00:06:12,093 pretty cool. 111 00:06:12,188 --> 00:06:12,818 Prasanna Malaiyandi: Point to go. 112 00:06:12,878 --> 00:06:13,238 Yeah. 113 00:06:13,238 --> 00:06:13,688 W. Curtis Preston: Yeah. 114 00:06:13,778 --> 00:06:15,398 Well, not just that, but also 115 00:06:15,578 --> 00:06:16,988 2.4 versus five. 116 00:06:17,048 --> 00:06:17,378 Yeah. 117 00:06:17,648 --> 00:06:19,868 Prasanna Malaiyandi: So actually all of this is part of the wifi standard, 118 00:06:19,868 --> 00:06:24,698 so the figuring out which access point, that's part of the 8 0 2 11 R standard. 119 00:06:25,148 --> 00:06:28,758 And I think that the band steering is also part of the standard as well. 120 00:06:29,648 --> 00:06:29,948 W. Curtis Preston: Yeah. 121 00:06:29,963 --> 00:06:31,583 Prasanna Malaiyandi: a lot of folks are implementing now. 122 00:06:31,763 --> 00:06:33,893 Some devices don't do well with band steering. 123 00:06:33,898 --> 00:06:36,413 It basically looks at sort of the difference between the five gigahertz 124 00:06:36,413 --> 00:06:39,473 and the 2.4 gigahertz and says, okay, which one should I pick? 125 00:06:39,923 --> 00:06:46,183 And most devices, if it's seven decibels difference or more, then it'll pick, 126 00:06:46,453 --> 00:06:48,853 uh, the higher the faster speed. 127 00:06:49,993 --> 00:06:53,083 And so that's kind of how it tricks your devices into picking the right band. 128 00:06:54,478 --> 00:06:55,138 W. Curtis Preston: Interesting. 129 00:06:55,858 --> 00:06:56,938 Yeah, it's kind of cool. 130 00:06:57,148 --> 00:07:04,768 Um, all I know is that I finally have a mesh that covers the two. 131 00:07:04,768 --> 00:07:09,598 Cuz my problem is that I have things in the garage, things embedded inside 132 00:07:09,598 --> 00:07:14,178 walls in the garage that need wifi, not just, not just inside walls. 133 00:07:14,998 --> 00:07:18,448 , I have a device that's inside a wall, inside an electrical 134 00:07:18,448 --> 00:07:19,708 cabinet, inside a wall. 135 00:07:20,128 --> 00:07:20,488 Right? 136 00:07:20,518 --> 00:07:26,278 I have a sense, uh, app or a bi, a device, and that's deep inside my 137 00:07:26,278 --> 00:07:29,278 electrical, my circuit breaker box. 138 00:07:29,908 --> 00:07:32,488 Um, and this reached to it. 139 00:07:32,488 --> 00:07:32,738 No problem. 140 00:07:33,078 --> 00:07:37,708 It didn't, it it had, it had like two bars. 141 00:07:38,068 --> 00:07:38,458 Right. 142 00:07:38,463 --> 00:07:41,738 So clearly, and, and the thing is, it's only, it's like 20 feet from. 143 00:07:42,218 --> 00:07:42,438 Prasanna Malaiyandi: yep. 144 00:07:43,033 --> 00:07:43,453 W. Curtis Preston: Right. 145 00:07:43,483 --> 00:07:48,103 But it's, you know, a couple of drywall walls and some 146 00:07:48,103 --> 00:07:49,993 two by fours and some metal. 147 00:07:50,653 --> 00:07:52,153 Uh, but it worked. 148 00:07:52,213 --> 00:07:53,983 That's the important part is that it worked. 149 00:07:54,703 --> 00:07:59,293 Um, yeah, so I th I think I might be in, I think I might 150 00:07:59,293 --> 00:08:00,673 be in wifi heaven for a while. 151 00:08:01,633 --> 00:08:08,473 Um, and you too can be there for the low, low price of $350 That's 152 00:08:08,478 --> 00:08:10,003 a two, that's a two node system. 153 00:08:10,753 --> 00:08:14,563 Um, And it's supposed like, yeah, but I'm pretty happy. 154 00:08:16,123 --> 00:08:18,283 But, uh, that's not what we're talking about today. 155 00:08:18,838 --> 00:08:19,093 Prasanna Malaiyandi: really. 156 00:08:19,098 --> 00:08:21,393 We can talk about wifi all day if you want. 157 00:08:21,553 --> 00:08:21,763 W. Curtis Preston: yeah. 158 00:08:21,763 --> 00:08:23,743 Well, you could talk about wifi all day. 159 00:08:23,743 --> 00:08:26,863 I feel really stupid when you're talking about wifi, because I'm 160 00:08:26,863 --> 00:08:30,673 like, this is not my Bailey Wick. 161 00:08:30,678 --> 00:08:33,043 That's a cool word, by the way, Bailey Wick. 162 00:08:34,663 --> 00:08:38,563 So I thought we'd talk about backups instead because that's, that's my world. 163 00:08:39,913 --> 00:08:41,173 And I feel comfortable knowing them. 164 00:08:41,178 --> 00:08:47,143 Most people don't know crap about this space, uh, because they, they, you know, 165 00:08:47,143 --> 00:08:51,793 they get the job as a junior person and then next thing you know, they become a, 166 00:08:52,033 --> 00:08:56,573 a real sys admin or a network admin or a, you know, or a security admin or a dba. 167 00:08:57,228 --> 00:08:59,638 Prasanna Malaiyandi: Yeah, well, except our listeners who are all 168 00:08:59,643 --> 00:09:02,668 awesome and probably experts in the backup field and know all about this. 169 00:09:02,733 --> 00:09:04,213 W. Curtis Preston: Well, certainly Daniel. 170 00:09:05,308 --> 00:09:05,698 Prasanna Malaiyandi: Hi Daniel. 171 00:09:06,298 --> 00:09:07,318 W. Curtis Preston: Hi Daniel. 172 00:09:07,323 --> 00:09:08,438 The backup anorak. 173 00:09:08,968 --> 00:09:13,708 Um, I wonder, you know, he's never, he's never, he better still be 174 00:09:13,708 --> 00:09:16,618 listening to the show since we call out to him every once in a while. 175 00:09:17,068 --> 00:09:18,688 Him and Stuart, although Stuart's retired. 176 00:09:19,538 --> 00:09:21,188 I don't think Stuart's listening to our show. 177 00:09:22,028 --> 00:09:23,498 I only tell 'em when we talk about 'em. 178 00:09:24,398 --> 00:09:28,658 But, um, so we're continuing in our backup to basic series. 179 00:09:28,658 --> 00:09:32,438 It's been a couple of weeks, uh, as the kids say it's been a minute, 180 00:09:33,098 --> 00:09:37,238 uh, since such a, I remember the first time I heard that thing, I was 181 00:09:37,238 --> 00:09:38,379 like, what are you talking a minute? 182 00:09:38,388 --> 00:09:43,583 Anyway, . But yeah, it's been a minute since we've done an episode of our 183 00:09:43,583 --> 00:09:48,593 Backup to Basic series, but I am looking down at the book and of course, 184 00:09:48,683 --> 00:09:53,363 uh, for those of you that don't know, basically we're doing a podcast version 185 00:09:53,393 --> 00:09:55,733 of my book, modern Data Protection. 186 00:09:55,733 --> 00:09:57,923 Make sure it gets in camera here from O'Reilly. 187 00:09:58,313 --> 00:10:02,603 Uh, you can purchase the, uh, the, the print version from, 188 00:10:02,693 --> 00:10:05,123 uh, your favorite book seller. 189 00:10:05,453 --> 00:10:13,548 Um, , perhaps it's one based in the Amazon, perhaps not, uh, Um, and, 190 00:10:13,548 --> 00:10:17,028 uh, but if you would like an ebook version of it, you can get your 191 00:10:17,028 --> 00:10:20,778 own by going to druva.com/ebook. 192 00:10:20,778 --> 00:10:23,418 That's d r uva.com/ebook. 193 00:10:23,928 --> 00:10:27,348 They will, of course, ask for your contact information and then email the crap 194 00:10:27,348 --> 00:10:28,758 out of you until you tell 'em to stop. 195 00:10:28,998 --> 00:10:31,838 But, That is, that is the price that you pay. 196 00:10:32,858 --> 00:10:34,778 Um, let's talk 197 00:10:34,778 --> 00:10:35,348 about, oh yeah. 198 00:10:36,008 --> 00:10:39,518 And while we're at it, uh, I'll throw out the disclaimer, uh, that 199 00:10:39,518 --> 00:10:43,978 this is an independent podcast and, um, uh, I work for Druva, 200 00:10:44,298 --> 00:10:47,328 Prasanna works for Zoom and, um, 201 00:10:48,388 --> 00:10:51,818 The, um, but the opinions that you hear are ours. 202 00:10:52,208 --> 00:10:53,498 Um, and. 203 00:10:54,548 --> 00:10:55,118 Et cetera. 204 00:10:55,438 --> 00:10:59,698 Please rate us, uh, by going to your, you know, most of you're on iTunes. 205 00:10:59,698 --> 00:11:04,888 Just scroll down to the bottom there, give us five or six stars and a comment. 206 00:11:04,888 --> 00:11:05,908 We love comments. 207 00:11:06,328 --> 00:11:11,528 And, uh, if you'd like to join the conversation, just contact me, w Curtis 208 00:11:11,938 --> 00:11:16,798 Preston gmail or WC Preston on Twitter. 209 00:11:17,583 --> 00:11:18,233 Prasanna Malaiyandi: What about LinkedIn? 210 00:11:19,078 --> 00:11:20,848 W. Curtis Preston: But n oh yeah, LinkedIn. 211 00:11:20,908 --> 00:11:23,188 Uh, it's linkedin.com/what is it? 212 00:11:23,188 --> 00:11:25,468 Slash in slash Mr. 213 00:11:25,468 --> 00:11:25,768 Beck. 214 00:11:26,578 --> 00:11:31,708 Um, and by the way, my Twitter account already has multifactor 215 00:11:31,708 --> 00:11:36,508 authentication, configured not using sms, which as should you, especially 216 00:11:36,508 --> 00:11:40,708 now that they're disabling, that so weird the way they did that. 217 00:11:40,858 --> 00:11:42,978 What's funny is I support the desysion. 218 00:11:43,198 --> 00:11:44,638 That's just the way 219 00:11:44,638 --> 00:11:44,788 they 220 00:11:44,878 --> 00:11:45,628 Prasanna Malaiyandi: way it came out. 221 00:11:45,658 --> 00:11:45,988 Yeah. 222 00:11:47,083 --> 00:11:48,103 W. Curtis Preston: Oh, Elon. 223 00:11:48,313 --> 00:11:48,733 Okay. 224 00:11:49,603 --> 00:11:54,823 So in our backup to basic series, we're continuing on, and today we are talking 225 00:11:54,823 --> 00:11:58,723 about using disk and deduplication. 226 00:11:59,833 --> 00:12:06,103 You know, I, I, um, couple weeks ago, I hit 30 years in the backup industry, 227 00:12:06,103 --> 00:12:08,463 and I got interviewed by Chris Mellor 228 00:12:09,403 --> 00:12:11,053 Prasanna Malaiyandi: the register and blocks and files. 229 00:12:12,103 --> 00:12:12,433 W. Curtis Preston: Yeah. 230 00:12:12,433 --> 00:12:14,663 It's in his, for his block and file. 231 00:12:15,388 --> 00:12:25,258 Um, um, blog and one of the questions was what I thought was the most, 232 00:12:26,308 --> 00:12:30,598 um, important development in the backup industry since I joined. 233 00:12:30,748 --> 00:12:37,978 And to me, hands down, it's not even, it's not, there's not even a close second, and 234 00:12:37,978 --> 00:12:41,008 that is the invention of deduplication 235 00:12:41,313 --> 00:12:41,533 Prasanna Malaiyandi: Yep. 236 00:12:42,088 --> 00:12:44,278 W. Curtis Preston: and because. 237 00:12:46,373 --> 00:12:56,923 I, I can't think of another technology in the backup space that has changed backup 238 00:12:56,923 --> 00:13:02,623 architecture more than deduplication, and I can think of many other things 239 00:13:02,623 --> 00:13:08,113 that we do that are only possible because deduplication is underneath them, 240 00:13:09,178 --> 00:13:10,678 Prasanna Malaiyandi: Oh yeah, definitely. 241 00:13:10,798 --> 00:13:10,978 Yeah. 242 00:13:10,978 --> 00:13:16,588 I don't think we would be able to get, especially with the data growth 243 00:13:16,768 --> 00:13:19,468 and the size of these applications. 244 00:13:19,648 --> 00:13:20,488 W. Curtis Preston: Is data growing? 245 00:13:20,698 --> 00:13:20,848 Is 246 00:13:20,968 --> 00:13:22,018 Prasanna Malaiyandi: No, not at all. 247 00:13:22,378 --> 00:13:22,678 Right. 248 00:13:22,683 --> 00:13:26,038 I don't think it would be possible to do, like I know Curtis, you've talked 249 00:13:26,038 --> 00:13:28,979 about previous, like in your early days, right, about trying to do a backup. 250 00:13:29,338 --> 00:13:32,218 I being like, oh my God, how am I gonna do this full backup in a weekend? 251 00:13:33,748 --> 00:13:34,318 W. Curtis Preston: Yeah. 252 00:13:34,363 --> 00:13:37,273 Prasanna Malaiyandi: And just with the fact, and I know we'll go and talk 253 00:13:37,273 --> 00:13:40,543 about more about deduplication, but yeah, just being able to now do that 254 00:13:40,543 --> 00:13:46,963 in a cost effective way, using new ways of actually doing the backups as well, 255 00:13:46,963 --> 00:13:48,313 which is enabled with deduplication. 256 00:13:48,838 --> 00:13:49,198 W. Curtis Preston: Yeah. 257 00:13:49,203 --> 00:13:52,198 So it, it's, it's like disk. 258 00:13:52,198 --> 00:13:58,258 You could argue that disk using disk and backups is the bigger, uh, advancement. 259 00:13:58,258 --> 00:14:00,508 But first off, not really an advancement. 260 00:14:00,508 --> 00:14:02,518 It's just instead of tape, we're gonna use disc, 261 00:14:03,138 --> 00:14:03,358 but 262 00:14:03,508 --> 00:14:04,798 Prasanna Malaiyandi: was there to start with anyway. 263 00:14:04,798 --> 00:14:09,988 It was just sort of, the cost was so high, and especially given the type of workload 264 00:14:09,988 --> 00:14:14,398 you see with deduplication where, or with backups where you're doing periodic 265 00:14:14,398 --> 00:14:18,328 fulls or other things like that, and keeping them for long periods of time. 266 00:14:18,333 --> 00:14:23,608 Are you going to spend what, 40 x or 30 x on storage for your backup 267 00:14:23,608 --> 00:14:25,018 system versus your production? 268 00:14:25,798 --> 00:14:26,158 Right. 269 00:14:26,218 --> 00:14:27,088 That's a hard sell. 270 00:14:27,133 --> 00:14:29,263 W. Curtis Preston: just, yeah, cuz that's a problem. 271 00:14:29,263 --> 00:14:33,403 So one of the, one of the things, uh, that I remember from back in the 272 00:14:33,403 --> 00:14:37,873 day, like I, I don't remember really thinking about this lately, but back 273 00:14:37,873 --> 00:14:42,538 in the day, I would say that for every gigabyte of primary storage, you 274 00:14:42,538 --> 00:14:45,028 had 20 gigabytes of backup storage. 275 00:14:45,868 --> 00:14:53,548 And so if you're gonna do that with disk, even, you know, even once, many years ago. 276 00:14:53,668 --> 00:14:54,088 Wow. 277 00:14:54,088 --> 00:14:59,068 At this point, it's like 20 years ago, . But, but even once they came out with 278 00:14:59,068 --> 00:15:03,298 this idea of, Uh, SATA disk instead 279 00:15:03,298 --> 00:15:03,508 of 280 00:15:03,568 --> 00:15:03,958 Prasanna Malaiyandi: nearline 281 00:15:03,958 --> 00:15:04,738 storage. 282 00:15:05,008 --> 00:15:05,488 W. Curtis Preston: Right. 283 00:15:05,908 --> 00:15:09,118 Um, that, that helped bring the cost down significantly. 284 00:15:09,598 --> 00:15:11,188 But, But, 285 00:15:11,188 --> 00:15:12,988 not, But not as much as deduplication. 286 00:15:13,273 --> 00:15:13,603 Prasanna Malaiyandi: Yeah. 287 00:15:13,633 --> 00:15:16,243 Because even with those price differences, right? 288 00:15:16,243 --> 00:15:19,783 Maybe it was half the price or a third of the price, but once you add 289 00:15:19,783 --> 00:15:23,593 in that 20 x that you talked about, right, Curtis, then that adds up. 290 00:15:23,983 --> 00:15:26,413 And it's not only just the storage cost, it's also you have 291 00:15:26,413 --> 00:15:30,493 to account for the power, the cooling, the floor space, right? 292 00:15:30,493 --> 00:15:32,163 All the things that go into that system. 293 00:15:33,373 --> 00:15:33,823 W. Curtis Preston: Yeah. 294 00:15:34,003 --> 00:15:34,423 Yeah. 295 00:15:34,933 --> 00:15:42,283 Um, it's funny, um, just sort of, just sort of a, an afterthought that, that. 296 00:15:43,393 --> 00:15:48,463 Post that, um, that Chris Mellor did about the 30 years. 297 00:15:48,913 --> 00:15:53,593 The one group that jumped on the article and just started retweeting all kinds of 298 00:15:53,683 --> 00:15:59,083 parts of, of, or pieces of the article was the tape group , because I said, 299 00:15:59,413 --> 00:16:01,273 I said really good things about tape. 300 00:16:01,693 --> 00:16:06,763 And, and the thing is that, um, you know, I, I, you know, I, I 301 00:16:06,763 --> 00:16:09,613 believe in all of those things, but. 302 00:16:11,383 --> 00:16:16,273 You know, all of the advancements that I've seen in backup in the last 20 plus 303 00:16:16,273 --> 00:16:18,943 years has been disk and deduplication. 304 00:16:19,723 --> 00:16:20,113 Right. 305 00:16:20,653 --> 00:16:26,413 Um, so let's talk about, so what, so not everybody really 306 00:16:26,418 --> 00:16:28,033 understands what deduplication is. 307 00:16:28,303 --> 00:16:34,963 Some people used to describe it like, well, it's like compression, uh, the way 308 00:16:34,963 --> 00:16:37,673 I remember it's like macro compression. 309 00:16:38,393 --> 00:16:41,003 Um, it's like compression over time. 310 00:16:41,723 --> 00:16:42,323 do you think of that? 311 00:16:44,303 --> 00:16:48,653 Prasanna Malaiyandi: uh, I don't quite like that, so, so, right. 312 00:16:49,048 --> 00:16:51,563 W. Curtis Preston: may be some old blog posts that I might have 313 00:16:51,683 --> 00:16:53,303 said that phrase, but go ahead. 314 00:16:53,723 --> 00:16:55,903 Prasanna Malaiyandi: so in my mind, right, deduplication is. 315 00:16:57,098 --> 00:17:02,978 Finding two identical segments and tossing one away, keeping only one copy, 316 00:17:03,038 --> 00:17:07,718 but still keeping a reference to that so you can, so you still know you have two 317 00:17:07,718 --> 00:17:09,548 virtual copies, but one physical copy, 318 00:17:10,178 --> 00:17:10,598 W. Curtis Preston: Mm-hmm. 319 00:17:11,168 --> 00:17:11,178 Prasanna Malaiyandi: right? 320 00:17:11,768 --> 00:17:13,358 At a high level, that's what I, and now 321 00:17:13,528 --> 00:17:13,648 W. Curtis Preston: you? 322 00:17:14,948 --> 00:17:21,338 Prasanna Malaiyandi: what is compression is taking an object, a singular object, 323 00:17:21,338 --> 00:17:23,228 and squeezing it into a smaller space. 324 00:17:24,368 --> 00:17:24,668 W. Curtis Preston: Right. 325 00:17:24,668 --> 00:17:27,188 But how do you understand how compression works? 326 00:17:27,188 --> 00:17:28,488 Cuz I Sure as hell don't 327 00:17:28,838 --> 00:17:30,488 Prasanna Malaiyandi: yeah, so typically like you would run it 328 00:17:30,488 --> 00:17:33,518 through different types of algorithms like LZ compression and all the rest 329 00:17:33,788 --> 00:17:39,998 in order to look for patterns and throw away bits and compress it down. 330 00:17:39,998 --> 00:17:42,188 Now, the difference I would say between duping compression 331 00:17:42,188 --> 00:17:43,358 because they do sound the same, 332 00:17:43,848 --> 00:17:44,138 W. Curtis Preston: Yeah. 333 00:17:44,158 --> 00:17:44,508 Prasanna Malaiyandi: right? 334 00:17:45,188 --> 00:17:50,178 I would say one of the differences is with deduplication. 335 00:17:52,583 --> 00:17:55,883 It's more like a file system level compression, if you want to 336 00:17:55,883 --> 00:17:58,583 think of it that way, because it's not just I'm taking this block. 337 00:17:59,363 --> 00:17:59,663 Yeah. 338 00:17:59,873 --> 00:18:02,633 It's not just I'm taking this and I'm squeezing it down such 339 00:18:02,638 --> 00:18:04,853 that it could be, I just need to look at this and figure it out. 340 00:18:04,853 --> 00:18:05,063 Right. 341 00:18:05,063 --> 00:18:07,793 It's a lot more complex than that. 342 00:18:07,913 --> 00:18:09,873 W. Curtis Preston: It is definitely a lot more complex than compression. 343 00:18:10,433 --> 00:18:10,793 Right. 344 00:18:11,183 --> 00:18:16,043 Um, I, I, I've just, I've, I've just honestly never really dug into the code 345 00:18:16,043 --> 00:18:18,713 of how traditional compression works. 346 00:18:19,113 --> 00:18:24,093 So the idea is that I'm looking for duplicate segments of data across many 347 00:18:24,093 --> 00:18:29,343 places, both from different sources as well as different time periods, right? 348 00:18:29,373 --> 00:18:33,303 I'm, I'm comparing the, this chunk of data that's coming in right 349 00:18:33,308 --> 00:18:35,673 now and, and tonight's backup. 350 00:18:35,733 --> 00:18:40,623 I'm comparing it literally with every chunk of data that I've ever received 351 00:18:41,583 --> 00:18:42,543 from anywhere else. 352 00:18:42,903 --> 00:18:45,543 . Prasanna Malaiyandi: I would say that's an ideal system, but not everyone 353 00:18:45,543 --> 00:18:47,103 builds their deduplication that way. 354 00:18:47,898 --> 00:18:48,438 , W. Curtis Preston: So, 355 00:18:48,663 --> 00:18:49,293 Prasanna Malaiyandi: where 356 00:18:49,563 --> 00:18:49,803 it's 357 00:18:49,938 --> 00:18:51,708 W. Curtis Preston: there are, yeah, go ahead. 358 00:18:52,233 --> 00:18:52,443 Prasanna Malaiyandi: Yeah. 359 00:18:52,443 --> 00:18:56,343 So it all goes down to sort of what is your deduplication domain is another 360 00:18:56,343 --> 00:18:58,353 term that some people talk about, right? 361 00:18:58,353 --> 00:19:00,663 Which is, is it limited to a system? 362 00:19:00,663 --> 00:19:04,353 Is it limited to a cluster which might be formed to multiple systems, 363 00:19:04,743 --> 00:19:08,613 or is it limited to sort of a single backup stream coming in? 364 00:19:09,663 --> 00:19:09,993 So 365 00:19:09,993 --> 00:19:10,263 there. 366 00:19:10,398 --> 00:19:12,618 W. Curtis Preston: that the question is what is your data domain? 367 00:19:12,658 --> 00:19:12,878 Uh, 368 00:19:13,518 --> 00:19:13,848 Prasanna Malaiyandi: Yeah. 369 00:19:14,898 --> 00:19:15,888 D Domain. 370 00:19:16,558 --> 00:19:17,518 W. Curtis Preston: So let's back up. 371 00:19:18,568 --> 00:19:22,528 So a, as I understand it, right, so basically we're taking the data 372 00:19:22,528 --> 00:19:26,008 that's, that's coming in or that's going to come in, we're slicing 373 00:19:26,008 --> 00:19:27,758 it up into, I like the term chunk. 374 00:19:28,603 --> 00:19:28,948 , right? 375 00:19:29,428 --> 00:19:33,298 We run those chunks through a cryptographic hashing algorithm. 376 00:19:33,838 --> 00:19:37,558 SH one, Shaw 2 56, whatever it, whatever you're using. 377 00:19:37,948 --> 00:19:43,658 On the other side of that, we get a alpha numeric value, in the case of SH one, 378 00:19:43,663 --> 00:19:46,898 it's 160 bit alpha alphanumeric value. 379 00:19:47,438 --> 00:19:54,038 so basically you, you, depending on the algorithm you use, you get a, um, you get 380 00:19:54,248 --> 00:19:59,738 an alpha numeric value at the end, and the size of that val, of that value is going 381 00:19:59,738 --> 00:20:02,348 to be based on which algorithm you use. 382 00:20:02,348 --> 00:20:06,128 In the case of SHA-1, it's 160 bits, right? 383 00:20:06,758 --> 00:20:07,148 And. 384 00:20:07,778 --> 00:20:10,538 You can then take the 160 bits. 385 00:20:10,538 --> 00:20:12,128 You can't reverse engineer it. 386 00:20:12,128 --> 00:20:17,288 You can't take the 160 bits and turn it into the chunk, but you can use that, that 387 00:20:17,288 --> 00:20:20,438 value to uniquely identify that chunk. 388 00:20:20,738 --> 00:20:24,008 And so if you have another chunk of data, regardless of where it came from, 389 00:20:25,598 --> 00:20:30,878 If it's 160 bit value, again, that's SHA-1 and other values are different. 390 00:20:31,088 --> 00:20:36,758 If it's fingerprint is the same, you can say that this chunk is identical 391 00:20:36,758 --> 00:20:40,118 to that other chunk that had the same fingerprint, and you can then 392 00:20:40,123 --> 00:20:43,008 discard the other chunk, right? 393 00:20:43,538 --> 00:20:43,688 the, 394 00:20:43,688 --> 00:20:43,928 the 395 00:20:44,123 --> 00:20:47,273 Prasanna Malaiyandi: Yeah, you can, you can discard the actual data, 396 00:20:47,278 --> 00:20:51,203 but you should still keep track of it somewhere in a file system, 397 00:20:51,203 --> 00:20:52,493 just because you need, still need 398 00:20:52,543 --> 00:20:52,723 W. Curtis Preston: Yeah. 399 00:20:52,723 --> 00:20:53,563 You're gonna keep track. 400 00:20:53,743 --> 00:20:55,363 Oh, we found another one of these, 401 00:20:55,543 --> 00:20:55,933 right? 402 00:20:57,013 --> 00:21:00,313 Prasanna Malaiyandi: And so usually that lookup is in a deduplication 403 00:21:00,343 --> 00:21:01,513 index is what they called them. 404 00:21:01,513 --> 00:21:04,783 Usually a dedupe index, which keeps a list of, Hey, here are 405 00:21:04,788 --> 00:21:06,158 all the fingerprints that I have. 406 00:21:06,748 --> 00:21:07,098 W. Curtis Preston: Right. 407 00:21:07,668 --> 00:21:12,528 As we, we were alluding to before, one of the things that determines 408 00:21:12,528 --> 00:21:17,478 sort of your effectiveness of, of dedupe is the dedupe domain, right? 409 00:21:17,478 --> 00:21:21,258 So I've seen it file system level, meaning it only looks for 410 00:21:21,263 --> 00:21:22,968 duplicate data within each volume. 411 00:21:23,868 --> 00:21:27,438 I've seen it host level, I've seen it backup level, meaning 412 00:21:27,438 --> 00:21:29,448 literally backup configuration wise. 413 00:21:30,288 --> 00:21:30,648 right? 414 00:21:30,648 --> 00:21:34,578 So if I, if I have a Windows server and I'm backing up the host and I'm 415 00:21:34,578 --> 00:21:39,198 backing up SQL Server, I only look for duplicates within SQL Server 416 00:21:39,228 --> 00:21:41,418 backups right against each other. 417 00:21:42,078 --> 00:21:49,148 Uh, then we have, um, if we're backing up several systems to a box, right? 418 00:21:49,598 --> 00:21:52,788 Maybe that the dedupe domain is only within that box. 419 00:21:52,788 --> 00:21:53,988 It's only looking for. 420 00:21:55,158 --> 00:21:56,688 Duplicates between all of that. 421 00:21:56,928 --> 00:22:02,218 And then there's what I would call truly global dedupe, which is , we're looking 422 00:22:02,218 --> 00:22:08,038 for duplicates from everything that's coming in, uh, from multiple sources. 423 00:22:08,098 --> 00:22:08,448 Right? 424 00:22:09,028 --> 00:22:09,448 Prasanna Malaiyandi: Mm-hmm. 425 00:22:10,348 --> 00:22:11,438 W. Curtis Preston: there is a. 426 00:22:12,703 --> 00:22:17,563 Point of decreasing marginal returns, right? 427 00:22:17,653 --> 00:22:21,803 You can argue, and certainly if you're a company that only does d dedupe within, 428 00:22:21,823 --> 00:22:25,123 like earlier I was, we only looked for dupes within SQL server backups. 429 00:22:25,753 --> 00:22:29,083 You could make an argument that, well, there's not a lot of duplicate data 430 00:22:29,088 --> 00:22:31,483 between SQL Server and Windows, right? 431 00:22:31,783 --> 00:22:35,623 so even though we're not comparing the two, there's not, there's not gonna 432 00:22:35,623 --> 00:22:38,413 be a lot of duplicate data there, and there's not gonna be a lot of duplicate. 433 00:22:39,573 --> 00:22:43,288 between the SQL Server database on this host and the SQL 434 00:22:43,288 --> 00:22:44,818 Server database on that host. 435 00:22:45,148 --> 00:22:46,708 So that's another argument that some 436 00:22:47,023 --> 00:22:50,083 Prasanna Malaiyandi: but, but I think a lot of that was because of 437 00:22:50,083 --> 00:22:54,133 architectural limitations of the products themselves rather than, 438 00:22:54,133 --> 00:22:55,843 that is really what you wanted to do. 439 00:22:55,843 --> 00:22:56,143 Right? 440 00:22:56,143 --> 00:22:56,443 Because 441 00:22:56,443 --> 00:22:57,643 that's more of a management issue. 442 00:22:58,378 --> 00:23:03,448 W. Curtis Preston: they didn't, It was like, it was like, well, if we're 443 00:23:03,448 --> 00:23:07,558 gonna do it, if we're gonna do it that way, it's gonna be much harder. 444 00:23:08,023 --> 00:23:10,843 to, to, to design a product to do it that way. 445 00:23:11,653 --> 00:23:13,963 And we don't think, we don't think that there's going to 446 00:23:13,963 --> 00:23:16,573 be that much more benefit, um, 447 00:23:17,308 --> 00:23:19,168 Prasanna Malaiyandi: But on the other hand, if you look 448 00:23:19,168 --> 00:23:20,638 at things like VMware, right? 449 00:23:20,638 --> 00:23:24,268 If I have a bunch of VMs, right, there's a good cha, and they all came 450 00:23:24,268 --> 00:23:25,948 from a single golden image, right? 451 00:23:25,948 --> 00:23:30,238 There's a good chance that as you're backing it up, 80, 90% of that stuff 452 00:23:30,238 --> 00:23:31,918 is all gonna be deduplicated, right? 453 00:23:32,563 --> 00:23:33,403 W. Curtis Preston: Absolutely. 454 00:23:33,613 --> 00:23:34,063 Yeah. 455 00:23:34,393 --> 00:23:39,913 There's also a lot of duplicate data even within like a large filer, right? 456 00:23:39,918 --> 00:23:41,983 There's gonna be lots of duplicate data there, right? 457 00:23:41,983 --> 00:23:45,823 So if you're only doing it volume to volume or backup configuration to 458 00:23:45,828 --> 00:23:48,613 backup configuration, you, there's a lot of duplicate data that I 459 00:23:48,613 --> 00:23:49,663 think you would, you would miss. 460 00:23:50,653 --> 00:23:51,283 Um, 461 00:23:52,093 --> 00:23:54,523 Prasanna Malaiyandi: I know you talked about the domains, but I think another 462 00:23:54,523 --> 00:23:59,983 thing to also mention is, Some products do different types of chunking, if you will. 463 00:24:00,163 --> 00:24:06,043 Some do it at the file level, others do it at sort of a smaller level, right? 464 00:24:06,048 --> 00:24:10,093 And some do sort of fixed segment where each one is sort of a fixed length. 465 00:24:10,423 --> 00:24:14,413 Others do sort of variable segments where they try to figure out what is optimal, 466 00:24:14,418 --> 00:24:18,343 because depending on how you're doing your fingerprinting, right, you want to 467 00:24:18,343 --> 00:24:20,623 find the most number of matches, right? 468 00:24:20,623 --> 00:24:22,093 So you can save on storage. 469 00:24:22,858 --> 00:24:23,278 W. Curtis Preston: right. 470 00:24:23,308 --> 00:24:23,518 I, 471 00:24:23,593 --> 00:24:24,493 Prasanna Malaiyandi: another thing that also comes up. 472 00:24:25,378 --> 00:24:27,838 W. Curtis Preston: I would argue that file level dedupe isn't really 473 00:24:27,838 --> 00:24:29,908 dedupe, it's more a single instance. 474 00:24:30,328 --> 00:24:30,838 Right. 475 00:24:30,868 --> 00:24:34,528 Um, that's like single instance storage of a file, you 476 00:24:34,528 --> 00:24:34,798 know? 477 00:24:35,518 --> 00:24:36,028 Okay. 478 00:24:36,238 --> 00:24:37,078 It, it's, yeah. 479 00:24:37,378 --> 00:24:42,868 But so I, I'm always thinking subfile, uh, when I think about what I think 480 00:24:42,868 --> 00:24:50,803 of actual dedupe . There is a much, like a very big, uh, other way that 481 00:24:51,093 --> 00:24:56,283 we divide up the dedupe industry, and that is source versus target. 482 00:24:56,793 --> 00:24:57,013 Prasanna Malaiyandi: Yep. 483 00:24:58,428 --> 00:25:06,798 W. Curtis Preston: Um, the, um, the first dedupe product I ever saw, 484 00:25:07,398 --> 00:25:13,458 which was, uh, no, was not, that was not the first, no, the first one I saw 485 00:25:13,463 --> 00:25:15,798 the product at the time was called Undo. 486 00:25:16,568 --> 00:25:17,328 Have we talked 487 00:25:17,328 --> 00:25:17,718 about this? 488 00:25:18,108 --> 00:25:18,328 Prasanna Malaiyandi: Mm. 489 00:25:19,218 --> 00:25:20,448 W. Curtis Preston: Undo with two Os. 490 00:25:20,478 --> 00:25:24,408 It was really funny that the name of a dedupe vendor. 491 00:25:25,743 --> 00:25:28,473 Had duplicate data in their company name. 492 00:25:28,953 --> 00:25:30,723 It was undoo with two os. 493 00:25:31,563 --> 00:25:34,833 You know this product, you just don't know that that's what it used to be called. 494 00:25:35,388 --> 00:25:35,748 Prasanna Malaiyandi: What is it? 495 00:25:37,938 --> 00:25:38,118 What 496 00:25:38,193 --> 00:25:41,163 W. Curtis Preston: give you a, I'll give, I'll give you a hint. 497 00:25:42,153 --> 00:25:42,483 It. 498 00:25:43,443 --> 00:25:48,993 The name comes from the fact that it would be a C of availability. 499 00:25:52,498 --> 00:25:54,748 I'm gonna, I'm gonna put the, the Jeopardy theme in here. 500 00:25:57,433 --> 00:26:01,063 Prasanna Malaiyandi: What would it see of availability? 501 00:26:01,683 --> 00:26:03,898 W. Curtis Preston: That's what the name, that's where the name for the company 502 00:26:03,898 --> 00:26:09,928 comes from, or if I want to put it in the right order, an availability c. 503 00:26:11,617 --> 00:26:13,087 Prasanna Malaiyandi: I don't know what this is. 504 00:26:14,467 --> 00:26:14,887 W. Curtis Preston: Avamar 505 00:26:15,657 --> 00:26:17,647 Prasanna Malaiyandi: Oh, oh, that makes sense. 506 00:26:18,577 --> 00:26:18,937 W. Curtis Preston: Yeah. 507 00:26:19,567 --> 00:26:21,397 So that's, that's where the name Avamar came from. 508 00:26:21,757 --> 00:26:22,867 So the, the first 509 00:26:23,137 --> 00:26:23,887 Prasanna Malaiyandi: I should know that 510 00:26:25,047 --> 00:26:25,867 W. Curtis Preston: you shouldn't know 511 00:26:25,972 --> 00:26:29,842 Prasanna Malaiyandi: I having being, uh, part of my former employer. 512 00:26:30,052 --> 00:26:30,412 Yes. 513 00:26:30,427 --> 00:26:30,937 W. Curtis Preston: Yeah. 514 00:26:31,117 --> 00:26:36,067 Well, I mean, you know, I, I have a bit of an inside track because that they're, They 515 00:26:36,067 --> 00:26:38,017 were right up the road from me, right? 516 00:26:38,017 --> 00:26:38,737 They were up there. 517 00:26:38,767 --> 00:26:39,607 They were up in Irvine. 518 00:26:40,327 --> 00:26:44,127 Um, and that was, uh, the first dedupe product. 519 00:26:44,157 --> 00:26:48,647 They were a source dedupe . So what's the difference between source 520 00:26:48,647 --> 00:26:50,457 dedupe and target dedupe Prasanna? 521 00:26:50,992 --> 00:26:54,352 Prasanna Malaiyandi: So the biggest one is, so let's first 522 00:26:54,352 --> 00:26:55,582 talk about target tup, right? 523 00:26:55,582 --> 00:27:00,832 So Target Tup is data comes into the system and then a deduplication 524 00:27:00,832 --> 00:27:02,902 algorithm runs tosses away data. 525 00:27:03,302 --> 00:27:06,122 It can support any type of client as long as it supports 526 00:27:06,127 --> 00:27:07,202 whatever the protocol it has. 527 00:27:07,222 --> 00:27:10,292 So it's NFS or smb, right? 528 00:27:10,652 --> 00:27:12,652 Whatever can write to it, the data gets deduped. 529 00:27:12,932 --> 00:27:13,472 W. Curtis Preston: Hang on, hang on. 530 00:27:13,472 --> 00:27:14,462 Before you go on to that. 531 00:27:14,722 --> 00:27:16,162 I don't disagree with what you said. 532 00:27:16,162 --> 00:27:18,562 I just, I think there could be a little bit more clarification. 533 00:27:18,952 --> 00:27:22,643 It's a box that I send whatever I want to. 534 00:27:23,592 --> 00:27:23,812 Prasanna Malaiyandi: Yep. 535 00:27:24,097 --> 00:27:27,667 W. Curtis Preston: Typically it, the thing about Target Dedup was that, 536 00:27:27,817 --> 00:27:31,087 um, that it was, you didn't have to do a lot of re-engineering of 537 00:27:31,087 --> 00:27:31,957 your backup system. 538 00:27:32,452 --> 00:27:34,042 Prasanna Malaiyandi: it's like a VTL system, right? 539 00:27:34,042 --> 00:27:34,252 That came. 540 00:27:34,357 --> 00:27:35,407 W. Curtis Preston: plug in a box. 541 00:27:35,467 --> 00:27:35,947 Yeah. 542 00:27:35,977 --> 00:27:40,267 And you would send you, and basically you stopped using tape and you 543 00:27:40,267 --> 00:27:41,357 sent your backups to this box. 544 00:27:41,647 --> 00:27:44,587 Maybe the box might even be pretending to be a tape library, 545 00:27:44,592 --> 00:27:45,847 the virtual tape library. 546 00:27:45,847 --> 00:27:46,147 Right. 547 00:27:46,777 --> 00:27:50,107 Um, and then it did all the dedupe magic over there. 548 00:27:50,677 --> 00:27:51,427 Um, 549 00:27:51,882 --> 00:27:53,512 Prasanna Malaiyandi: Which was great because you can 550 00:27:53,512 --> 00:27:54,712 just plug in your box and go. 551 00:27:55,687 --> 00:27:59,137 Now the other side is called source side dedupe, instead of sending all the 552 00:27:59,137 --> 00:28:02,497 data and tossing it away, why don't we do something smart and actually figure 553 00:28:02,497 --> 00:28:08,107 out the duplicates on the client itself, on the source right, dedupe on the 554 00:28:08,112 --> 00:28:09,857 source, and only send the unique data. 555 00:28:10,237 --> 00:28:12,607 And this has the advantage. 556 00:28:13,792 --> 00:28:16,732 Actually not sending the data over the wire, which is actually 557 00:28:16,732 --> 00:28:20,482 a huge benefit that people don't understand always, right? 558 00:28:20,482 --> 00:28:24,322 Is not sending the data can actually make it a lot faster, even though 559 00:28:24,322 --> 00:28:27,692 you think, oh, I'm now putting additional load on my server itself. 560 00:28:27,692 --> 00:28:32,972 But it ends up being better than trying to send all the data and just tossing 561 00:28:32,972 --> 00:28:34,832 it away like target-side dedupe does. 562 00:28:35,327 --> 00:28:37,667 W. Curtis Preston: I would say it theoretically should be better 563 00:28:38,747 --> 00:28:39,137 right? 564 00:28:39,377 --> 00:28:43,817 Because you, I'm just saying I've seen some crappy source dedupe systems, right? 565 00:28:43,877 --> 00:28:44,177 Prasanna Malaiyandi: Okay. 566 00:28:44,387 --> 00:28:44,597 Sorry. 567 00:28:44,597 --> 00:28:47,357 I've seen some, I've seen some good ones, or the ones that I've 568 00:28:47,357 --> 00:28:48,737 interacted with have been good. 569 00:28:48,737 --> 00:28:51,707 And so I've seen the performance numbers around 570 00:28:52,292 --> 00:28:52,622 W. Curtis Preston: Yeah. 571 00:28:52,622 --> 00:28:55,862 I, I do think it, it makes more sense to me. 572 00:28:55,982 --> 00:28:57,632 It always made more sense to me. 573 00:28:57,842 --> 00:29:02,772 The only reason why we had Target dedupe was because to do source dedupe , you 574 00:29:02,772 --> 00:29:03,992 have to redesign the backup product. 575 00:29:04,982 --> 00:29:05,372 , right? 576 00:29:05,462 --> 00:29:09,572 It took a long time to get, to get, uh, basically you have to 577 00:29:09,572 --> 00:29:13,742 stop using net backup networker or tsm, whatever it was back in the 578 00:29:13,747 --> 00:29:15,602 day, and you had to replace it. 579 00:29:15,632 --> 00:29:19,952 Like in this case with Avamar, Avamar was a source do-do product. 580 00:29:19,952 --> 00:29:21,752 You had to do what we call a four clipped upgrade. 581 00:29:21,752 --> 00:29:23,922 You had to throw out the baby with the bathwater, whatever phrase, whatever. 582 00:29:25,247 --> 00:29:25,667 You know, 583 00:29:26,267 --> 00:29:28,427 uh, analogy you want to use there. 584 00:29:28,877 --> 00:29:32,387 That was the main problem as I saw it with source dedup. 585 00:29:32,447 --> 00:29:32,777 Right. 586 00:29:32,777 --> 00:29:37,667 Is that, is that you, you had to change your backup product to get it, 587 00:29:38,507 --> 00:29:40,307 Prasanna Malaiyandi: and that was in the beginning, right? 588 00:29:40,787 --> 00:29:41,687 At the very early 589 00:29:41,707 --> 00:29:44,327 W. Curtis Preston: Well, well, You. 590 00:29:44,327 --> 00:29:45,137 You, well, yeah. 591 00:29:45,137 --> 00:29:49,067 Now you just had to, had to upgrade your backup product, right? 592 00:29:49,067 --> 00:29:55,727 Because many of modern backup technologies now support source dedupe , although 593 00:29:56,417 --> 00:30:01,277 even some newer backup technologies don't, I don't know if, I dunno if that 594 00:30:01,277 --> 00:30:05,597 came out in English, so some I, there was some double negatives in there. 595 00:30:06,257 --> 00:30:11,597 Some newer, very new backup technologies. 596 00:30:12,197 --> 00:30:14,007 Don't do source dedupe 597 00:30:14,364 --> 00:30:15,534 . Prasanna Malaiyandi: which seems bunkers. 598 00:30:15,734 --> 00:30:17,294 W. Curtis Preston: which does seem bonkers. 599 00:30:17,414 --> 00:30:22,544 Um, I, you know, and, um, I'm talking about the likes of 600 00:30:22,549 --> 00:30:24,254 Rubric and Cohesity, right? 601 00:30:24,259 --> 00:30:29,594 These are new, these are, you know, next gen backup products that were designed 602 00:30:29,594 --> 00:30:32,744 in the last, less than the last 10 years. 603 00:30:33,314 --> 00:30:33,734 Right. 604 00:30:34,454 --> 00:30:36,524 And it's based on an appliance model. 605 00:30:37,264 --> 00:30:42,019 and they do all the dedupe inside that box, is my understanding, right? 606 00:30:43,289 --> 00:30:46,379 Prasanna Malaiyandi: And I just wanna challenge that, Curtis, because I thought 607 00:30:46,379 --> 00:30:53,759 in some cases, They do do source side deduplication, but I think because they've 608 00:30:53,759 --> 00:30:58,139 tried to be open and act as a target device, in those cases, you can't, like, 609 00:30:58,144 --> 00:30:59,489 you don't really have another option. 610 00:31:00,254 --> 00:31:03,584 W. Curtis Preston: Yeah, I, I don't, well, again, I'm not, 611 00:31:03,599 --> 00:31:04,229 Prasanna Malaiyandi: I, but I don't know 612 00:31:04,279 --> 00:31:08,444 W. Curtis Preston: work at, I work at Druva, not at Rubrik, uh, or, or Cohesity. 613 00:31:08,444 --> 00:31:12,584 But it is my understanding that they do target side dedup, which is, and, 614 00:31:12,674 --> 00:31:16,924 and one of the challenges of target side dedup is you need an appliance. 615 00:31:18,114 --> 00:31:19,469 at each location. 616 00:31:19,949 --> 00:31:22,619 Now I know that they can do virtual appliances, right? 617 00:31:22,619 --> 00:31:24,899 So they have a, they have a VM level appliance. 618 00:31:25,349 --> 00:31:30,119 Uh, but you need a box or something pretending to be a box at each location, 619 00:31:30,449 --> 00:31:34,919 because if you're not eliminating the duplicates before you send it 620 00:31:34,924 --> 00:31:40,199 to the box, um, then you need, you need something that's on-prem, right? 621 00:31:40,694 --> 00:31:41,624 Prasanna Malaiyandi: Because you definitely don't wanna 622 00:31:41,624 --> 00:31:42,634 send that all over the Wan 623 00:31:43,259 --> 00:31:46,669 W. Curtis Preston: No, no, that's the, to me, that's the biggest advantage 624 00:31:46,669 --> 00:31:50,809 of a source dedupe system is that it's ultimately scalable, right? 625 00:31:50,809 --> 00:31:54,889 That you, that assuming, assuming it doesn't slow things down, assuming, 626 00:31:55,429 --> 00:31:59,619 assuming all these things, assuming that the product actually works, um, that 627 00:32:00,649 --> 00:32:04,429 you, um, you could back up a laptop. 628 00:32:05,159 --> 00:32:05,524 , right? 629 00:32:05,524 --> 00:32:09,964 You can back up a mobile phone and the, the duplicate data will be eliminated 630 00:32:09,964 --> 00:32:12,934 before it's sent over the wan, which is what you need to do if you're 631 00:32:12,939 --> 00:32:14,404 backing up something over the internet. 632 00:32:15,124 --> 00:32:15,484 Prasanna Malaiyandi: Mm-hmm. 633 00:32:15,994 --> 00:32:16,384 W. Curtis Preston: Right. 634 00:32:16,714 --> 00:32:24,484 Um, and, um, so the, the downside that some, you know, again, you, 635 00:32:24,484 --> 00:32:28,384 you, you talked about it already, is that it does put additional 636 00:32:28,714 --> 00:32:31,864 compute requirement on the client. 637 00:32:34,114 --> 00:32:37,234 The argument is that it's offset by the, 638 00:32:37,684 --> 00:32:40,774 um, the savings of the network bandwidth. 639 00:32:40,774 --> 00:32:41,104 Right. 640 00:32:41,644 --> 00:32:42,304 Um, 641 00:32:42,934 --> 00:32:44,194 Prasanna Malaiyandi: There is also one more downside, 642 00:32:44,944 --> 00:32:45,094 W. Curtis Preston: okay. 643 00:32:45,274 --> 00:32:46,324 Prasanna Malaiyandi: which is that. 644 00:32:47,224 --> 00:32:51,514 Not all applications can do source side deduplication. 645 00:32:52,084 --> 00:32:57,004 So if you do have an application which only supports writing to like 646 00:32:57,004 --> 00:33:01,744 an NFS Mount point or an SMB Mount point, or something that doesn't 647 00:33:01,749 --> 00:33:05,484 allow the integration of these source side deduplication duplication logic, 648 00:33:05,484 --> 00:33:08,704 then you are going to need to be able to support target side dedupe. 649 00:33:08,704 --> 00:33:08,754 do. 650 00:33:09,624 --> 00:33:10,074 W. Curtis Preston: Yep. 651 00:33:10,584 --> 00:33:11,484 Uh, agreed. 652 00:33:11,874 --> 00:33:17,814 Um, and an example of that would be like, um, uh, Oracle, right? 653 00:33:17,934 --> 00:33:18,294 Prasanna Malaiyandi: Yep. 654 00:33:18,834 --> 00:33:20,034 Incremental merge. 655 00:33:21,264 --> 00:33:21,684 W. Curtis Preston: yeah. 656 00:33:22,044 --> 00:33:27,024 Um, although I would think that you should be able, I don't know, we could, we 657 00:33:27,084 --> 00:33:27,784 Prasanna Malaiyandi: No, you can't. 658 00:33:27,784 --> 00:33:28,224 You can't. 659 00:33:28,324 --> 00:33:28,784 You can't. 660 00:33:29,244 --> 00:33:32,634 W. Curtis Preston: You can't take the Oracle stream and slice it and dice it. 661 00:33:36,504 --> 00:33:37,104 I don't know. 662 00:33:37,119 --> 00:33:37,779 Prasanna Malaiyandi: Did you what? 663 00:33:37,989 --> 00:33:38,289 Sorry? 664 00:33:38,439 --> 00:33:43,509 You could, um, there are companies out there which give, which provide 665 00:33:43,509 --> 00:33:45,429 a virtual file system interface 666 00:33:46,119 --> 00:33:46,779 that lives 667 00:33:46,854 --> 00:33:47,904 W. Curtis Preston: So you you fake it. 668 00:33:47,904 --> 00:33:48,654 You fake it out. 669 00:33:49,104 --> 00:33:49,344 Yeah. 670 00:33:49,344 --> 00:33:49,584 Okay. 671 00:33:50,964 --> 00:33:51,354 All right. 672 00:33:51,354 --> 00:33:54,744 And then I've got something called hybrid dedupe and this, this was 673 00:33:54,749 --> 00:33:56,454 invented by your former employer. 674 00:33:58,409 --> 00:34:00,299 Prasanna Malaiyandi: I don't even know what a hybrid dedupe is. 675 00:34:01,464 --> 00:34:04,524 W. Curtis Preston: it's, it's, it's Target Dedoo pretending to be Source cdu. 676 00:34:08,124 --> 00:34:08,334 It's 677 00:34:08,499 --> 00:34:08,679 Prasanna Malaiyandi: D. 678 00:34:09,729 --> 00:34:17,589 Oh, see, here's my, okay, so here's my problem is I think Boost 679 00:34:18,054 --> 00:34:18,404 W. Curtis Preston: Uhhuh. 680 00:34:19,239 --> 00:34:22,509 Prasanna Malaiyandi: is source. 681 00:34:22,509 --> 00:34:27,249 I deduplication, I don't know if I would call it hybrid, because it is 682 00:34:27,254 --> 00:34:29,289 very similar to what Avamar DI did. 683 00:34:30,569 --> 00:34:30,864 , right? 684 00:34:30,869 --> 00:34:34,404 It's moving the deduplication logic to the client 685 00:34:35,364 --> 00:34:37,854 such that you could do all of the computation. 686 00:34:37,854 --> 00:34:41,124 The same thing that we have talked about with source I deduplication, 687 00:34:41,889 --> 00:34:44,409 W. Curtis Preston: I, I'll tell you why I put it in a different category. 688 00:34:44,989 --> 00:34:48,279 To me, hybrid dedupe is redoing the backup software. 689 00:34:49,119 --> 00:34:52,509 I'm sorry, source dedupe, true source iDation. 690 00:34:53,619 --> 00:34:55,479 It's done at the backup software level, 691 00:34:55,974 --> 00:34:56,784 Prasanna Malaiyandi: Okay, then. 692 00:34:56,784 --> 00:34:56,874 I 693 00:34:56,949 --> 00:35:01,449 W. Curtis Preston: with, with with hybrid dedupe . I'm still dumb sending 694 00:35:01,569 --> 00:35:05,619 everything to this source dedupe thing that's gonna redo it, right? 695 00:35:05,679 --> 00:35:11,379 Um, it doesn't matter in the end, you get, you get roughly the same benefits, right? 696 00:35:11,839 --> 00:35:13,929 Um, that's what, uh, 697 00:35:13,989 --> 00:35:14,319 Prasanna Malaiyandi: Okay. 698 00:35:14,319 --> 00:35:15,189 So with hybrid, yeah. 699 00:35:15,189 --> 00:35:19,509 You get the benefits of source without having to upgrade and, or sorry, 700 00:35:19,514 --> 00:35:20,829 throw away your backup software. 701 00:35:21,429 --> 00:35:22,869 W. Curtis Preston: Right, right, right. 702 00:35:23,469 --> 00:35:30,939 Um, so I, I, um, we spent most of this time talking about dedupe . Um, 703 00:35:30,999 --> 00:35:34,299 there are a bunch of different ways to use disk in your backup system. 704 00:35:35,349 --> 00:35:37,659 Some of which don't really require dedup, right? 705 00:35:37,659 --> 00:35:40,749 We used to do what we call disk cashing, where you just had enough 706 00:35:40,749 --> 00:35:42,219 disk for last night's backup. 707 00:35:42,219 --> 00:35:45,729 You would back up to disk and then you would copy that to tape, and then 708 00:35:45,729 --> 00:35:47,229 you would hand that to a man in a van. 709 00:35:47,679 --> 00:35:49,269 Uh, then we got a bunch of different things. 710 00:35:49,269 --> 00:35:53,169 I got D to D to T D to D to D, D to D, D to C, and D to D to to C. 711 00:35:53,669 --> 00:35:54,609 Did I do all that? 712 00:35:54,969 --> 00:35:59,289 So dis to dis to tape disc, to disc to disk, direct cloud and 713 00:35:59,294 --> 00:36:00,399 dis to disc to cloud, right? 714 00:36:00,399 --> 00:36:06,159 So these are all ways that people use disk in current backup systems. 715 00:36:06,669 --> 00:36:11,770 Um, to me, d D to C or disto disc to cloud is really dis to disc. 716 00:36:11,949 --> 00:36:13,629 To disc is just the cloud is or the 717 00:36:13,629 --> 00:36:16,029 disc Is being run by the cloud, right? 718 00:36:16,419 --> 00:36:20,109 And I will say that dedupe , by the way, I will say that without d. 719 00:36:22,104 --> 00:36:27,084 The whole thing of using the cloud, the way we use the cloud just wouldn't work. 720 00:36:28,404 --> 00:36:30,414 I mean, you can't send full backups to the cloud. 721 00:36:30,419 --> 00:36:33,414 I mean, you could, with unlimited bandwidth. 722 00:36:34,089 --> 00:36:36,309 Prasanna Malaiyandi: well, and yeah, with unlimited bandwidth 723 00:36:36,314 --> 00:36:37,629 it would just be expensive. 724 00:36:37,629 --> 00:36:37,779 Right. 725 00:36:37,779 --> 00:36:40,029 Just going back to the conversation we had earlier about the wan, right? 726 00:36:40,479 --> 00:36:43,929 You don't wanna send full copies out to over the wan. 727 00:36:44,694 --> 00:36:45,044 W. Curtis Preston: right. 728 00:36:45,474 --> 00:36:48,654 Prasanna Malaiyandi: Um, because that gets expensive and very slow. 729 00:36:49,074 --> 00:36:54,654 Um, the other one I was going to comment on was, uh, oh, I know we've 730 00:36:54,654 --> 00:36:58,434 been talking about disk, but I think it's also important to acknowledge 731 00:36:58,434 --> 00:37:02,034 that now it's no longer spinning disk. 732 00:37:02,034 --> 00:37:03,144 It could also be flash. 733 00:37:03,924 --> 00:37:04,274 Right. 734 00:37:04,564 --> 00:37:05,424 We've seen 735 00:37:06,054 --> 00:37:06,594 W. Curtis Preston: yeah, 736 00:37:06,594 --> 00:37:08,034 but that's a whole other thing 737 00:37:08,234 --> 00:37:09,954 Prasanna Malaiyandi: I I, I, know, but I'm just saying that when it 738 00:37:09,954 --> 00:37:15,039 comes to deduplication and backup ST or protection storage, right? 739 00:37:15,459 --> 00:37:19,899 This, it could be flash, it could be disk, it could be object storage, right? 740 00:37:20,649 --> 00:37:23,649 So I think it's important to differentiate that, like what we're 741 00:37:23,649 --> 00:37:26,439 talking about with deduplication, when we mentioned disk, right? 742 00:37:26,439 --> 00:37:27,909 The media layer itself. 743 00:37:28,689 --> 00:37:29,979 Yeah, the media layer. 744 00:37:30,009 --> 00:37:30,429 Yes. 745 00:37:30,819 --> 00:37:32,320 The media layer is not tape. 746 00:37:33,989 --> 00:37:35,019 W. Curtis Preston: Right, right. 747 00:37:35,619 --> 00:37:36,669 Hang on one second. 748 00:37:37,119 --> 00:37:47,994 Um, I need to, didn't realize I had a, I had a, um, Meeting 749 00:37:48,084 --> 00:37:48,444 Prasanna Malaiyandi: Meaning a. 750 00:37:50,184 --> 00:37:50,664 W. Curtis Preston: Yeah. 751 00:37:50,724 --> 00:37:51,144 Four. 752 00:37:51,144 --> 00:38:07,484 Well, four 15, which is an odd, um, all right. 753 00:38:11,349 --> 00:38:14,649 It's a, it's a pre-meeting with a podcast thing. 754 00:38:14,649 --> 00:38:24,699 It's, um, anyway, um, so, uh, yeah, so, okay, you know, I hate the idea of flash 755 00:38:24,804 --> 00:38:25,824 Prasanna Malaiyandi: know, I know, I know. 756 00:38:25,824 --> 00:38:29,244 I'm, I, I'm just saying that people will bring it up. 757 00:38:29,249 --> 00:38:32,064 So I just wanna clarify that when we talk about disc, we're 758 00:38:32,064 --> 00:38:33,234 just talking about not tape. 759 00:38:34,089 --> 00:38:35,769 W. Curtis Preston: The only place. 760 00:38:36,009 --> 00:38:36,879 Yeah. 761 00:38:36,909 --> 00:38:37,479 Correct. 762 00:38:37,809 --> 00:38:43,689 The only place where I think maybe Flash has a place in the backup 763 00:38:43,694 --> 00:38:48,369 system is, and you know, you know, the folks over at Pierre and 764 00:38:48,369 --> 00:38:50,049 Neil, they're all mad at me now. 765 00:38:50,049 --> 00:38:50,349 Right. 766 00:38:50,799 --> 00:38:56,559 But, uh, the only place that I, where I think Flash has a place in the backup 767 00:38:56,559 --> 00:38:59,439 system is with like live recovery. 768 00:39:01,404 --> 00:39:05,094 If you're gonna do, if you're gonna do instant recovery and you're actually gonna 769 00:39:05,094 --> 00:39:10,704 run VMs off of your backups, that better be some really nice performing disk. 770 00:39:11,274 --> 00:39:13,914 But the thing is, it doesn't need to be your whole system. 771 00:39:13,914 --> 00:39:15,144 It just needs to be like the most 772 00:39:15,369 --> 00:39:17,949 Prasanna Malaiyandi: A part, part of, and it needs to, you don't need 773 00:39:17,954 --> 00:39:19,419 your entire system to be flash, 774 00:39:20,124 --> 00:39:20,414 W. Curtis Preston: Yeah. 775 00:39:20,439 --> 00:39:20,799 Prasanna Malaiyandi: You just 776 00:39:20,804 --> 00:39:23,380 need enough to be able to support that use case. 777 00:39:24,024 --> 00:39:28,524 W. Curtis Preston: I, I just think that where Flash does really, 778 00:39:28,524 --> 00:39:32,229 really, Is in random access, right? 779 00:39:32,679 --> 00:39:35,049 Backup isn't a random access application. 780 00:39:36,339 --> 00:39:37,899 Backup is a streaming application. 781 00:39:37,899 --> 00:39:40,559 Even if what we're talking is large dedupe chunks. 782 00:39:40,779 --> 00:39:41,019 I don't 783 00:39:41,019 --> 00:39:41,199 know. 784 00:39:41,439 --> 00:39:41,769 I, I, 785 00:39:41,769 --> 00:39:41,889 Prasanna Malaiyandi: I, 786 00:39:41,979 --> 00:39:42,129 I, 787 00:39:42,129 --> 00:39:44,709 W. Curtis Preston: say, let's just say the jury is out for me. 788 00:39:44,709 --> 00:39:46,119 I, I am in Missouri. 789 00:39:46,149 --> 00:39:46,719 Missouri. 790 00:39:47,019 --> 00:39:49,119 Is that, is that the show me state? 791 00:39:49,119 --> 00:39:49,989 That's the show me state. 792 00:39:49,989 --> 00:39:49,990 Right? 793 00:39:50,229 --> 00:39:50,649 Prasanna Malaiyandi: yeah. 794 00:39:51,849 --> 00:39:53,379 W. Curtis Preston: So I'll tell you what, I'll tell you what. 795 00:39:53,739 --> 00:39:57,729 If there's anybody that's listening to this that just got pissed off, 796 00:39:59,334 --> 00:39:59,724 Prasanna Malaiyandi: what's his 797 00:39:59,724 --> 00:39:59,934 name? 798 00:39:59,934 --> 00:40:00,504 I'll come back 799 00:40:00,609 --> 00:40:03,759 W. Curtis Preston: to, I welcome you to, come on and tell me why I'm wrong. 800 00:40:03,789 --> 00:40:04,599 I, I just, 801 00:40:04,664 --> 00:40:07,014 Prasanna Malaiyandi: I, I, I, I know who will come back on, you 802 00:40:07,089 --> 00:40:07,539 W. Curtis Preston: who, who, 803 00:40:07,539 --> 00:40:08,139 will come back on, 804 00:40:08,444 --> 00:40:09,104 Prasanna Malaiyandi: what's his name? 805 00:40:09,214 --> 00:40:10,024 Bass Data guy. 806 00:40:11,379 --> 00:40:11,839 W. Curtis Preston: uh oh. 807 00:40:12,019 --> 00:40:14,739 Oh, are they flash 808 00:40:15,604 --> 00:40:15,894 Prasanna Malaiyandi: Yeah, 809 00:40:16,569 --> 00:40:17,049 W. Curtis Preston: mark? 810 00:40:18,009 --> 00:40:20,049 Um, No, sorry, Howard. 811 00:40:20,649 --> 00:40:21,309 Uh, Howard. 812 00:40:21,939 --> 00:40:22,389 Yeah. 813 00:40:22,744 --> 00:40:23,234 Prasanna Malaiyandi: Fastest. 814 00:40:23,514 --> 00:40:24,174 Pure flash. 815 00:40:24,624 --> 00:40:24,864 Yeah. 816 00:40:25,329 --> 00:40:25,809 W. Curtis Preston: Yeah. 817 00:40:25,869 --> 00:40:27,849 Um, all right. 818 00:40:27,939 --> 00:40:28,389 All right. 819 00:40:28,509 --> 00:40:33,099 Well, yeah, Howard, uh, you wanna tell, you wanna tell me why I'm wrong? 820 00:40:33,159 --> 00:40:35,169 Um, I'm more than happy to have you back. 821 00:40:35,529 --> 00:40:36,399 We can duke it out. 822 00:40:36,489 --> 00:40:36,999 We can duke it. 823 00:40:37,359 --> 00:40:38,410 wouldn't be the first time. 824 00:40:38,415 --> 00:40:40,689 Howard and I have, have disagreed on something. 825 00:40:40,929 --> 00:40:41,259 I don't know. 826 00:40:41,259 --> 00:40:44,289 It's just, it's just there are so many area, there are so many 827 00:40:44,289 --> 00:40:47,399 other places where I would wanna spend money in the backup system. 828 00:40:48,774 --> 00:40:48,894 Prasanna Malaiyandi: Yep. 829 00:40:48,939 --> 00:40:52,329 W. Curtis Preston: Um, and, um, 830 00:40:52,464 --> 00:40:54,054 Prasanna Malaiyandi: comes down to what the cost is. 831 00:40:54,054 --> 00:40:54,264 Right. 832 00:40:54,264 --> 00:40:56,904 If you could get flash down to a low enough point, 833 00:40:57,309 --> 00:41:00,249 W. Curtis Preston: which is the point of vast data, right? 834 00:41:00,309 --> 00:41:06,309 Their architecture allows using flash in a, um, you know, a significant way, 835 00:41:06,444 --> 00:41:07,314 Prasanna Malaiyandi: That's, that's why I brought 836 00:41:07,419 --> 00:41:08,159 W. Curtis Preston: uh, close to cost. 837 00:41:08,409 --> 00:41:08,739 Okay. 838 00:41:08,949 --> 00:41:09,249 All right. 839 00:41:09,309 --> 00:41:09,399 Okay. 840 00:41:10,029 --> 00:41:10,270 All right. 841 00:41:11,259 --> 00:41:11,979 All right. 842 00:41:12,489 --> 00:41:13,269 All right. 843 00:41:13,839 --> 00:41:17,529 Um, and then I got this whole other thing. 844 00:41:17,529 --> 00:41:18,759 I'm not gonna go into that other thing. 845 00:41:19,329 --> 00:41:29,769 Um, but yeah, so d d makes disk and, and cloud-based products, both physiologically 846 00:41:29,769 --> 00:41:33,039 feasible as well as economically feasible. 847 00:41:33,489 --> 00:41:33,909 Right. 848 00:41:34,449 --> 00:41:35,229 Um, 849 00:41:35,679 --> 00:41:36,009 Prasanna Malaiyandi: is. 850 00:41:37,419 --> 00:41:37,749 W. Curtis Preston: hmm. 851 00:41:37,974 --> 00:41:41,544 Prasanna Malaiyandi: Is there something that a person shopping for 852 00:41:41,544 --> 00:41:45,294 a dedupe system should be asking? 853 00:41:46,314 --> 00:41:48,324 Like what are the important things that they should be 854 00:41:48,324 --> 00:41:50,064 asking in order to determine 855 00:41:50,274 --> 00:41:52,374 W. Curtis Preston: yeah, that's a, that's a great question. 856 00:41:52,404 --> 00:42:00,534 I think the, the question would be things about what's the restored performance? 857 00:42:00,534 --> 00:42:02,514 Because in the end, that's the only thing that matters. 858 00:42:02,964 --> 00:42:03,864 I remember. 859 00:42:04,944 --> 00:42:05,754 A product. 860 00:42:06,204 --> 00:42:10,254 Now, this product is still on the market, but I believe, I believe 861 00:42:10,254 --> 00:42:12,894 they have addressed this, this issue. 862 00:42:13,464 --> 00:42:15,444 I remember a dedupe product. 863 00:42:15,504 --> 00:42:21,384 It was a Target dedupe product that had, uh, I remember that had 400 megabytes 864 00:42:21,444 --> 00:42:25,074 a second throughput in to an appliance. 865 00:42:25,554 --> 00:42:26,604 Prasanna Malaiyandi: And like 10 megabits out 866 00:42:27,144 --> 00:42:31,974 W. Curtis Preston: It was 40, it was 40, it was 40, uh, megabytes out. 867 00:42:32,304 --> 00:42:35,194 It had a 90%, what we call dedupe tax. 868 00:42:35,214 --> 00:42:35,694 Right. 869 00:42:36,174 --> 00:42:39,564 That the, because the problem with dedupe, depending on how 870 00:42:39,564 --> 00:42:42,684 you store it, is that you've got everything you need all over the 871 00:42:42,954 --> 00:42:43,914 Prasanna Malaiyandi: All over the place. 872 00:42:43,974 --> 00:42:44,264 W. Curtis Preston: Yeah. 873 00:42:44,694 --> 00:42:48,174 And so this was just a really, really, really bad design. 874 00:42:48,834 --> 00:42:56,004 And um, uh, I believe that they addressed it and, um, because that 875 00:42:56,004 --> 00:42:57,474 product is still on the market today. 876 00:42:57,474 --> 00:43:00,894 But that version, one of that product was ble. 877 00:43:01,734 --> 00:43:04,914 Um, so yeah, it's about restored performance, right? 878 00:43:04,914 --> 00:43:07,244 So one thing, oh, I'm. 879 00:43:07,544 --> 00:43:10,594 Uh, dedupe ratio is crap. 880 00:43:10,924 --> 00:43:12,394 Don't look at dedupe ratio. 881 00:43:12,394 --> 00:43:14,529 dedupe ratio is a made up number. 882 00:43:14,949 --> 00:43:19,229 Um, I will, um, I'll, I'll go back to, I'll pick on Avamar. 883 00:43:19,234 --> 00:43:19,779 Avamar. 884 00:43:19,784 --> 00:43:23,889 Back in the day, they used to say they had a 400 to one DEDUP ratio. 885 00:43:23,979 --> 00:43:24,759 Do you remember this? 886 00:43:24,764 --> 00:43:25,149 Because 887 00:43:25,389 --> 00:43:28,449 they basically considered every backup as a full backup. 888 00:43:28,449 --> 00:43:31,299 They're like, the way we store backups, which is the same way Druva stores 889 00:43:31,299 --> 00:43:33,309 backups, the way we store backups. 890 00:43:33,399 --> 00:43:36,369 It's like, even though they're incremental, it's like they're a full. 891 00:43:36,924 --> 00:43:37,194 , right? 892 00:43:37,199 --> 00:43:39,834 Because they behave like a full during a restore. 893 00:43:40,254 --> 00:43:42,564 And so they considered every backup a full. 894 00:43:42,714 --> 00:43:46,074 And so they said, well then therefore the dedup ratio is 400 to one. 895 00:43:46,074 --> 00:43:47,844 Well, that was always complete nonsense. 896 00:43:48,384 --> 00:43:52,914 Um, the other would be, I remember, uh, again, I'm gonna pick on people equally. 897 00:43:52,919 --> 00:43:55,614 I remember sales reps of a certain large target. 898 00:43:56,139 --> 00:44:01,029 D company that where you might've worked, where they would tell customers to go 899 00:44:01,029 --> 00:44:04,629 and do full backups more frequently because it made their dedup ratio better. 900 00:44:05,619 --> 00:44:08,049 , which is just, again, nonsense. 901 00:44:08,409 --> 00:44:16,399 What matters, in my opinion, what matters is how big is a full backup versus 902 00:44:16,869 --> 00:44:19,689 how big are all the backups, right? 903 00:44:20,259 --> 00:44:21,609 So if I have. 904 00:44:22,254 --> 00:44:24,714 If I, let me, let me explain what I'm saying. 905 00:44:25,044 --> 00:44:29,574 If I have a hundred terabytes, if, if one full backup of my environment is a hundred 906 00:44:29,574 --> 00:44:35,574 terabytes and then after three months how big is, or whatever number you want. 907 00:44:36,294 --> 00:44:41,424 Uh, but it's just three months seems like a, a nice, long, um, what do you call it? 908 00:44:41,424 --> 00:44:43,734 Uh, POC thing, 909 00:44:43,734 --> 00:44:44,094 right? 910 00:44:45,144 --> 00:44:49,794 Um, after a hundred, after, you know, three months, how. 911 00:44:50,964 --> 00:44:53,274 How much stuff is stored over there? 912 00:44:53,964 --> 00:44:54,804 That's what I'm saying. 913 00:44:54,804 --> 00:44:59,354 Don't dedupe ratios is nonsense that that didn't come out in English. 914 00:44:59,514 --> 00:45:06,114 dedupe ratios are nonsense, but if I can fit a hundred terabytes right, if I have 915 00:45:06,119 --> 00:45:10,034 a hundred terabyte environment and then a series of incremental backups, and then 916 00:45:10,374 --> 00:45:12,804 over there, my question is how big is. 917 00:45:14,334 --> 00:45:16,884 How much data did I write to disk? 918 00:45:17,394 --> 00:45:23,274 And let's say it's, it's, it's 200 terabytes after 90 days. 919 00:45:23,484 --> 00:45:27,354 And then compare that with another product who writes a hundred terabytes? 920 00:45:27,684 --> 00:45:32,455 You backed up the same data, but you used half as much storage on the back end. 921 00:45:33,039 --> 00:45:33,819 . That's what I'm saying. 922 00:45:34,089 --> 00:45:37,059 The the, the problem is, and the, the other reason, and again, 923 00:45:37,089 --> 00:45:39,939 I'm a little extra sensitive to this cuz I work for Druva. 924 00:45:40,089 --> 00:45:42,289 People ask us what's our, what's our dedupe ratio? 925 00:45:42,309 --> 00:45:45,579 We're like, well the thing is we're like the opposite of Avamar. 926 00:45:45,609 --> 00:45:49,449 Well we're actually similar to Avamar in that we're source I dedupe, 927 00:45:49,929 --> 00:45:51,589 but we don't use that funny math. 928 00:45:51,689 --> 00:45:54,879 So we could say 401, but that's nonsense. 929 00:45:55,209 --> 00:45:57,879 So you know, we say, well, we. 930 00:45:58,704 --> 00:46:01,854 Because, because we also do incremental forever backups. 931 00:46:02,214 --> 00:46:03,444 That's, that's the problem. 932 00:46:03,444 --> 00:46:03,774 Right. 933 00:46:03,774 --> 00:46:09,504 So, um, but I know that on average, if we have a hundred terabyte 934 00:46:09,504 --> 00:46:15,384 customer, we store, you know, roughly a year's worth of backups in less 935 00:46:15,384 --> 00:46:16,824 than a hundred terabytes of disk. 936 00:46:16,924 --> 00:46:17,139 Prasanna Malaiyandi: Yeah. 937 00:46:17,559 --> 00:46:22,479 And I think it's important there to also account for that increment, like 938 00:46:22,659 --> 00:46:25,059 how I look at these like numbers. 939 00:46:25,119 --> 00:46:28,179 I totally get what you said, Curtis, like you should just do an apples apples. 940 00:46:28,179 --> 00:46:31,329 But if you don't have that ability, you should also look to say, okay, 941 00:46:31,809 --> 00:46:33,519 I have a hundred terabyte full. 942 00:46:33,729 --> 00:46:35,560 And then say, my daily change rate is 2%. 943 00:46:36,679 --> 00:46:36,894 right? 944 00:46:36,894 --> 00:46:39,024 So if I do 2% for a month, right? 945 00:46:39,024 --> 00:46:41,634 That's, what is that two 60? 946 00:46:42,654 --> 00:46:44,034 60 more terabytes, right? 947 00:46:44,034 --> 00:46:48,384 So it should be 160 terabytes worth of data that I sent over, right? 948 00:46:48,384 --> 00:46:51,834 For 160 terabytes worth of data, how much should I actually store? 949 00:46:53,184 --> 00:46:53,394 Right? 950 00:46:53,394 --> 00:46:55,604 Which will give you similar things to what you're saying, right? 951 00:46:56,104 --> 00:46:59,724 But Bec, because what you're saying is if you had the two products, then 952 00:46:59,729 --> 00:47:00,894 you could do a direct comparison. 953 00:47:00,894 --> 00:47:03,264 But I'm saying if you don't have the two products, then 954 00:47:03,264 --> 00:47:04,494 here's another way you could 955 00:47:04,524 --> 00:47:06,804 W. Curtis Preston: Well, I, well, I would argue that there's no way 956 00:47:06,804 --> 00:47:11,124 to compare them if you don't have two pro, if you, if you're not, if 957 00:47:11,124 --> 00:47:13,344 you're not doing a true comparison. 958 00:47:13,404 --> 00:47:13,914 Right. 959 00:47:14,389 --> 00:47:14,509 Prasanna Malaiyandi: A 960 00:47:14,574 --> 00:47:17,214 W. Curtis Preston: it's just, it's just that d math is funny, right? 961 00:47:18,429 --> 00:47:20,829 So different products charge differently, right? 962 00:47:20,829 --> 00:47:25,509 You look at, um, like when you look at Metallic, which competes 963 00:47:25,509 --> 00:47:29,139 with Druva, they have a frontend price and we have a backend price. 964 00:47:29,409 --> 00:47:32,139 They have, they actually have the front end price, and then you also 965 00:47:32,139 --> 00:47:33,669 need to pay for the backend storage. 966 00:47:33,849 --> 00:47:34,199 Right? 967 00:47:34,689 --> 00:47:37,119 So you're paying, so how do you, how do you compare that? 968 00:47:37,839 --> 00:47:41,589 Um, it's, it's just, it's difficult 969 00:47:41,634 --> 00:47:42,144 Prasanna Malaiyandi: hard. 970 00:47:42,204 --> 00:47:42,564 Yeah. 971 00:47:42,699 --> 00:47:43,479 W. Curtis Preston: it's hard. 972 00:47:43,839 --> 00:47:49,359 Uh, but all I'm saying is dedup ratio is crap and doesn't mean anything. 973 00:47:49,809 --> 00:47:53,979 Um, but what does matter is how much data are you storing on that 974 00:47:53,979 --> 00:47:56,950 backend because you will be paying for that one way or the other. 975 00:47:58,374 --> 00:47:58,854 All right. 976 00:47:59,604 --> 00:48:04,704 I don't know if we made this, if we, if this is clear as mud or what, but, uh, I 977 00:48:04,704 --> 00:48:09,834 hope that was helpful and, uh, maybe we, maybe we ticked off Howard and Howard's 978 00:48:09,834 --> 00:48:11,994 gonna come on next week's episode. 979 00:48:12,174 --> 00:48:12,534 . I dunno. 980 00:48:15,039 --> 00:48:15,639 Prasanna Malaiyandi: Come join us, 981 00:48:15,639 --> 00:48:16,269 Howard, 982 00:48:16,294 --> 00:48:19,014 W. Curtis Preston: Thanks for, thanks for, uh, thanks for helping 983 00:48:19,014 --> 00:48:20,904 me with my network as well, so, 984 00:48:21,249 --> 00:48:21,939 Prasanna Malaiyandi: anytime, Curtis. 985 00:48:22,749 --> 00:48:24,699 Just remember I am not tech support. 986 00:48:26,184 --> 00:48:26,589 W. Curtis Preston: Yeah. 987 00:48:26,769 --> 00:48:27,189 Yeah. 988 00:48:27,729 --> 00:48:28,269 All right. 989 00:48:28,479 --> 00:48:31,899 Well, uh, and thanks to the listeners and remember to subscribe 990 00:48:31,899 --> 00:48:33,789 so that you can restore it all.