1 00:00:00,000 --> 00:00:00,509 Yeah. 2 00:00:00,614 --> 00:00:03,104 the other thing people don't realize is five gigahertz, 3 00:00:03,494 --> 00:00:05,654 like degrades like no tomorrow 4 00:00:06,629 --> 00:00:08,279 me why five gertz is better again. 5 00:00:09,450 --> 00:00:14,280 It's faster because it can handle more bandwidth, and also the channel 6 00:00:14,280 --> 00:00:17,970 is wider, so you can have more things talking at the same time. 7 00:00:17,970 --> 00:00:21,060 It's just as your frequency goes up, the distance goes 8 00:00:21,060 --> 00:00:23,070 down for the same power levels, 9 00:00:23,429 --> 00:00:25,409 So is this like DC versus ac? 10 00:00:25,980 --> 00:00:27,540 not quite DC versus ac. 11 00:00:27,870 --> 00:00:28,389 It's more about. 12 00:00:29,534 --> 00:00:33,404 You need to pump as many things as possible into, because high frequency, 13 00:00:33,404 --> 00:00:38,964 it's more per cycle, than 2.4, which is less airtime, if you will. 14 00:00:39,744 --> 00:00:43,765 And so every sort of peak, you can send more out with the five gigahertz 15 00:00:43,769 --> 00:00:45,894 because you're doing it more often. 16 00:00:46,765 --> 00:00:49,105 And so it works a lot better. 17 00:00:49,105 --> 00:00:51,445 It's just the distance isn't as great. 18 00:00:51,805 --> 00:00:54,415 Now, I will tell people, so this is one of my, I'm gonna 19 00:00:54,420 --> 00:00:56,215 get up on my soapbox now, right? 20 00:00:56,365 --> 00:01:00,205 One of my rare soapbox events and tell people, a lot of times people 21 00:01:00,205 --> 00:01:05,245 think they need more wifi access points in their house to get coverage. 22 00:01:05,355 --> 00:01:05,775 Mm-hmm. 23 00:01:06,065 --> 00:01:08,965 to those people, I will say, plan out your network carefully. 24 00:01:08,965 --> 00:01:10,705 Put your devices where they matter. 25 00:01:11,075 --> 00:01:11,495 Mm-hmm. 26 00:01:11,560 --> 00:01:16,450 And also don't put too many devices and don't crank up the power all the way 27 00:01:16,455 --> 00:01:19,510 to high, because I know Curtis, you and I were talking about this when you're 28 00:01:19,510 --> 00:01:23,350 looking at mesh, and it was like, imagine that your router can overpower your 29 00:01:23,350 --> 00:01:29,080 phone, your laptop, your iPad, so it's screaming at the top of its lungs and your 30 00:01:29,085 --> 00:01:31,510 phone can barely even scream back at it. 31 00:01:31,510 --> 00:01:35,560 And so that's actually worse for your network and for airtime than 32 00:01:35,560 --> 00:01:37,059 actually balancing out power. 33 00:01:37,720 --> 00:01:41,059 I just don't know if, the stuff you're talking about, like is. 34 00:01:42,504 --> 00:01:46,634 is that even, is that configuration option even on consumer class routers? 35 00:01:47,028 --> 00:01:49,239 you'll have the low, medium, high power levels, 36 00:01:50,039 --> 00:01:52,949 but it takes time to fine tune and tweet these, right? 37 00:01:52,954 --> 00:01:57,299 You have to walk around with a wifi analyzer on your phone, right? 38 00:01:57,299 --> 00:01:59,989 So Apple with their, iPhones, right? 39 00:01:59,989 --> 00:02:01,399 They ship, what is it? 40 00:02:01,399 --> 00:02:03,549 Airport utility, which has a wifi scan. 41 00:02:04,239 --> 00:02:06,699 Option, which will show you all the wifi networks and the signal 42 00:02:06,699 --> 00:02:09,159 strength, and you basically have to walk around your house with that and 43 00:02:09,159 --> 00:02:10,299 be like, okay, where is it strong? 44 00:02:10,299 --> 00:02:10,989 Where is it weak? 45 00:02:11,899 --> 00:02:12,769 figure out the placement. 46 00:02:12,799 --> 00:02:16,189 That's the ideal way, because what you want is you want coverage in the right 47 00:02:16,189 --> 00:02:20,924 places, because what is in a lot of high density housing areas, or even homes next 48 00:02:20,929 --> 00:02:25,604 to each other is most people end up with crummy wifi because their power is turned 49 00:02:25,604 --> 00:02:31,034 up so high, it bleeds into everyone else's area such that everyone has a crappy time. 50 00:02:31,664 --> 00:02:34,509 because then you get interference and then everyone slows down and then it 51 00:02:34,609 --> 00:02:34,819 Yeah. 52 00:02:34,819 --> 00:02:36,019 I got a lot of wifi. 53 00:02:36,289 --> 00:02:37,519 I got a lot of networks. 54 00:02:38,239 --> 00:02:38,629 yeah, 55 00:02:39,289 --> 00:02:44,119 and for the last bit of my soapbox is please do not use 40 megahertz channel 56 00:02:44,119 --> 00:02:46,459 widths on your 2.4 gigahertz channels. 57 00:02:46,699 --> 00:02:51,439 You do not need to use 40 megahertz and ruin everyone else's connectivity. 58 00:02:51,859 --> 00:02:55,429 Please only use 20 megahertz bands for 2.4 gigahertz. 59 00:02:56,834 --> 00:02:57,609 I'll see what I can do 60 00:03:00,169 --> 00:03:03,889 but I have this new, and again, I am not a wireless, I feel like a 61 00:03:03,919 --> 00:03:09,019 wireless nbe, but I have this new fancy right where it automatically selects 62 00:03:09,054 --> 00:03:09,474 Mm-hmm. 63 00:03:09,619 --> 00:03:10,189 the right. 64 00:03:10,329 --> 00:03:10,749 that's 65 00:03:10,764 --> 00:03:11,124 access. 66 00:03:11,124 --> 00:03:11,754 Point to go. 67 00:03:11,814 --> 00:03:12,174 Yeah. 68 00:03:12,174 --> 00:03:12,624 Yeah. 69 00:03:12,824 --> 00:03:13,324 And actually 70 00:03:13,394 --> 00:03:14,054 also 71 00:03:14,264 --> 00:03:14,484 the. 72 00:03:14,894 --> 00:03:15,224 versus 73 00:03:15,404 --> 00:03:15,734 Yep. 74 00:03:16,304 --> 00:03:19,874 So actually all of this is part of the wifi standard, so the figuring 75 00:03:19,874 --> 00:03:23,354 out which access point, that's part of the 8 0 2 11 R standard. 76 00:03:23,804 --> 00:03:27,414 And I think that the band steering is also part of the standard as well. 77 00:03:28,304 --> 00:03:28,604 Yeah. 78 00:03:28,619 --> 00:03:30,239 a lot of folks are implementing now. 79 00:03:30,419 --> 00:03:32,549 Some devices don't do well with band steering. 80 00:03:32,554 --> 00:03:35,069 It basically looks at sort of the difference between the five gigahertz 81 00:03:35,069 --> 00:03:38,129 and the 2.4 gigahertz and says, okay, which one should I pick? 82 00:03:38,579 --> 00:03:44,239 And most devices, if it's seven decibels difference or more, then 83 00:03:44,239 --> 00:03:46,959 it'll pick, the higher the faster speed. 84 00:03:48,099 --> 00:03:50,919 And so that's how it tricks your devices into picking the right band. 85 00:03:52,314 --> 00:03:52,974 Interesting. 86 00:03:53,694 --> 00:03:54,534 Yeah, it's cool. 87 00:03:54,774 --> 00:03:55,704 It's magic though. 88 00:03:55,854 --> 00:04:01,194 that I finally have mesh that covers the two. 89 00:04:01,194 --> 00:04:05,634 Cuz my problem is that I have things in the garage, things embedded 90 00:04:05,634 --> 00:04:09,974 inside walls in the garage that need wifi, not just inside walls. 91 00:04:10,794 --> 00:04:14,244 , I have a device that's inside a wall, inside an electrical 92 00:04:14,244 --> 00:04:15,504 cabinet, inside a wall. 93 00:04:15,964 --> 00:04:17,934 have a sense, app or 94 00:04:17,934 --> 00:04:18,264 Yep. 95 00:04:18,324 --> 00:04:23,394 a device, and that's deep inside my electrical, my circuit breaker 96 00:04:23,414 --> 00:04:23,634 No. 97 00:04:24,614 --> 00:04:26,504 and this reached to it. 98 00:04:26,504 --> 00:04:26,754 No problem. 99 00:04:27,094 --> 00:04:29,174 It didn't, it had two bars. 100 00:04:29,484 --> 00:04:32,339 So clearly, and the thing is, it's only, it's 20 feet from. 101 00:04:32,819 --> 00:04:33,039 yep. 102 00:04:33,584 --> 00:04:36,724 But it's, a couple of drywall walls and 103 00:04:36,739 --> 00:04:36,979 And, 104 00:04:37,054 --> 00:04:38,764 by fours and some metal. 105 00:04:39,384 --> 00:04:40,014 but it worked. 106 00:04:40,074 --> 00:04:41,844 That's the important part is that it worked. 107 00:04:42,564 --> 00:04:46,384 yeah, so I th I think I might be in, I think I might be in wifi heaven 108 00:04:47,049 --> 00:04:47,739 Woohoo. 109 00:04:48,084 --> 00:04:54,264 and you too can be there for the low price of $350 That's a 110 00:04:54,264 --> 00:04:55,734 two, that's a two node system. 111 00:04:56,694 --> 00:04:58,914 it's supposed yeah, but I'm pretty happy. 112 00:05:00,474 --> 00:05:02,274 But, that's not what we're talking about today. 113 00:05:02,829 --> 00:05:03,084 really. 114 00:05:03,089 --> 00:05:05,384 We can talk about wifi all day if you want. 115 00:05:05,544 --> 00:05:05,754 yeah. 116 00:05:05,754 --> 00:05:07,404 you could talk about wifi all day. 117 00:05:07,404 --> 00:05:10,344 I feel really stupid when you're talking about wifi, because 118 00:05:10,349 --> 00:05:14,334 I'm like, this is not my Wick. 119 00:05:14,339 --> 00:05:16,704 That's a cool word, by the way, Bailey Wick. 120 00:05:18,324 --> 00:05:21,529 So I thought we'd talk about backups instead because that's my world. 121 00:05:22,879 --> 00:05:24,139 And I feel comfortable knowing them. 122 00:05:24,144 --> 00:05:29,059 Most people don't know crap about this space, because they, they get 123 00:05:29,064 --> 00:05:33,369 the job as a junior person and then next thing you know, they become a 124 00:05:33,369 --> 00:05:37,434 real sys admin or a network admin or a, or a security admin or a dba. 125 00:05:38,089 --> 00:05:41,679 Yeah, except our listeners who are all awesome and probably experts in the 126 00:05:41,679 --> 00:05:43,269 backup field and know all about this. 127 00:05:43,374 --> 00:05:44,304 certainly Daniel. 128 00:05:45,399 --> 00:05:45,789 Hi Daniel. 129 00:05:46,389 --> 00:05:47,409 Hi Daniel. 130 00:05:47,414 --> 00:05:48,529 The backup anorak. 131 00:05:49,059 --> 00:05:53,229 I wonder, he's never, he better still be listening to the show since we 132 00:05:53,234 --> 00:05:54,669 call out to him every once in a while. 133 00:05:55,119 --> 00:05:56,739 Him and Stuart, although Stuart's retired. 134 00:05:57,589 --> 00:05:59,239 I don't think Stuart's listening to our show. 135 00:06:00,079 --> 00:06:01,549 I only tell 'em when we talk about 'em. 136 00:06:02,449 --> 00:06:06,089 But, so we're continuing in our backup to basic series. 137 00:06:06,089 --> 00:06:08,959 It's been a couple of weeks, as the kids say it's been a minute, 138 00:06:09,569 --> 00:06:13,289 since a, I remember the first time I heard that thing, I was like, 139 00:06:13,294 --> 00:06:14,220 what are you talking a minute? 140 00:06:14,229 --> 00:06:19,424 Anyway, . But yeah, it's been a minute since we've done an episode of our 141 00:06:19,424 --> 00:06:24,604 Backup to Basic series, I am looking down at the book and of course, for 142 00:06:24,604 --> 00:06:29,044 those of you that don't know, basically we're doing a podcast version of 143 00:06:29,049 --> 00:06:31,114 my book, modern Data Protection. 144 00:06:31,114 --> 00:06:33,304 Make sure it gets in camera here from O'Reilly. 145 00:06:33,644 --> 00:06:39,079 you can purchase the, the print version from, your favorite seller. 146 00:06:39,409 --> 00:06:44,660 , perhaps it's one based in the Amazon, perhaps not, and, but if you would like 147 00:06:44,660 --> 00:06:50,600 an ebook version of it, you can get your own by going to druva.com/ebook. 148 00:06:50,600 --> 00:06:53,240 That's d r uva.com/ebook. 149 00:06:53,900 --> 00:06:57,320 will, of course, ask for your contact information then email the crap out 150 00:06:57,320 --> 00:06:58,580 of you until you tell 'em to stop. 151 00:06:58,820 --> 00:07:01,060 But, that is the price that you pay. 152 00:07:02,080 --> 00:07:02,830 let's talk 153 00:07:02,830 --> 00:07:03,220 about, oh 154 00:07:03,310 --> 00:07:03,530 And 155 00:07:04,060 --> 00:07:04,720 And while we're at 156 00:07:04,950 --> 00:07:09,750 I'll throw out the disclaimer, that this is an independent podcast I work 157 00:07:09,755 --> 00:07:13,320 for Druva, Prasanna works for Zoom and, 158 00:07:13,420 --> 00:07:15,810 The, but the opinions that you hear are ours. 159 00:07:16,200 --> 00:07:16,710 and. 160 00:07:17,760 --> 00:07:18,330 Et cetera. 161 00:07:18,650 --> 00:07:22,260 Please rate us, by going to your, most of you're on iTunes. 162 00:07:22,260 --> 00:07:27,450 Just scroll down to the bottom there, give us five or six stars and a comment. 163 00:07:27,450 --> 00:07:28,470 We love comments. 164 00:07:28,890 --> 00:07:33,710 And, if you'd like to join the conversation, just contact me, w Curtis 165 00:07:34,120 --> 00:07:38,980 Preston gmail or WC Preston on Twitter. 166 00:07:39,765 --> 00:07:40,415 What about LinkedIn? 167 00:07:41,260 --> 00:07:43,030 But n yeah, LinkedIn. 168 00:07:43,080 --> 00:07:45,000 it's linkedin.com/what is it? 169 00:07:45,000 --> 00:07:45,810 Slash in 170 00:07:45,995 --> 00:07:46,215 Yep. 171 00:07:46,770 --> 00:07:47,580 Mr. Beck. 172 00:07:48,390 --> 00:07:53,730 and by the way, my Twitter account already has multifactor authentication, configured 173 00:07:53,735 --> 00:07:58,580 not using sms, which as should you, especially now that they're disabling, 174 00:07:58,580 --> 00:08:01,620 that so weird the way they did that. 175 00:08:01,770 --> 00:08:03,890 What's funny is I support the desysion. 176 00:08:04,110 --> 00:08:05,550 That's just way 177 00:08:05,670 --> 00:08:06,030 the way 178 00:08:06,030 --> 00:08:06,540 it came out. 179 00:08:06,570 --> 00:08:06,900 Yeah. 180 00:08:07,995 --> 00:08:09,015 Oh, Elon. 181 00:08:09,225 --> 00:08:09,645 Okay. 182 00:08:10,515 --> 00:08:15,735 So in our backup to basic series, we're continuing on, and today we are talking 183 00:08:15,735 --> 00:08:19,635 about using disk and deduplication. 184 00:08:20,695 --> 00:08:25,936 I, I, couple weeks ago, I hit 30 years in the backup industry, and 185 00:08:25,936 --> 00:08:28,295 I got interviewed by Chris Mellor 186 00:08:29,235 --> 00:08:30,885 the register and blocks and files. 187 00:08:31,935 --> 00:08:32,265 Yeah. 188 00:08:32,265 --> 00:08:34,495 It's in his, for his block and file. 189 00:08:35,170 --> 00:08:35,740 blog 190 00:08:38,350 --> 00:08:45,540 one of the questions was what I thought was the most, important development 191 00:08:45,570 --> 00:08:48,000 in the backup industry since I joined. 192 00:08:48,150 --> 00:08:55,380 And to me, hands down, not even, not, there's not even a close second, and 193 00:08:55,380 --> 00:08:58,410 that is the invention of deduplication 194 00:08:58,715 --> 00:08:58,935 Yep. 195 00:08:59,490 --> 00:09:01,680 and because. 196 00:09:03,725 --> 00:09:14,105 I can't think of another technology in the backup space that has changed backup 197 00:09:14,105 --> 00:09:19,805 architecture more than deduplication, and I can think of many other things 198 00:09:19,805 --> 00:09:25,295 that we do that are only possible because deduplication is underneath them, 199 00:09:26,360 --> 00:09:27,860 Oh yeah, definitely. 200 00:09:27,980 --> 00:09:28,160 Yeah. 201 00:09:28,160 --> 00:09:33,770 I don't think we would be able to get, especially with the data growth 202 00:09:33,950 --> 00:09:36,650 and the size of these applications. 203 00:09:36,830 --> 00:09:37,670 Is data growing? 204 00:09:37,880 --> 00:09:38,030 Is 205 00:09:38,150 --> 00:09:39,200 No, not at all. 206 00:09:39,510 --> 00:09:42,865 I don't think it would be possible to do, like I know Curtis, you've talked 207 00:09:42,865 --> 00:09:45,576 about previous, like in your early days, about trying to do a backup. 208 00:09:45,935 --> 00:09:48,815 I being like, oh my God, how am I gonna do this full backup in a weekend? 209 00:09:50,345 --> 00:09:50,915 Yeah. 210 00:09:50,960 --> 00:09:54,410 And just with the fact, and I know we'll go and talk about more about 211 00:09:54,410 --> 00:09:58,040 deduplication, but yeah, just being able to now do that in a cost 212 00:09:58,040 --> 00:10:03,560 effective way, using new ways of actually doing the backups as well, 213 00:10:03,560 --> 00:10:04,910 which is enabled with deduplication. 214 00:10:05,435 --> 00:10:05,795 Yeah. 215 00:10:05,800 --> 00:10:07,945 So it's like disk. 216 00:10:07,945 --> 00:10:13,695 You could argue that disk using disk and backups the bigger, advancement. 217 00:10:13,695 --> 00:10:15,945 But first off, not really an advancement. 218 00:10:15,945 --> 00:10:17,955 It's just instead of tape, we're gonna use disc, 219 00:10:18,780 --> 00:10:20,235 which was there to start with anyway. 220 00:10:20,235 --> 00:10:25,125 It was just the cost was so high, and especially given the type of workload 221 00:10:25,125 --> 00:10:29,205 with deduplication where, or with backups where you're doing periodic 222 00:10:29,205 --> 00:10:33,135 fulls or other things like that, and keeping them for long periods of time. 223 00:10:33,140 --> 00:10:38,415 Are you going to spend what, 40 x or 30 x on storage for your backup 224 00:10:38,415 --> 00:10:39,825 system versus your production? 225 00:10:40,565 --> 00:10:41,435 That's a hard sell. 226 00:10:41,480 --> 00:10:43,610 just, yeah, cuz that's a problem. 227 00:10:43,610 --> 00:10:47,070 So one of the, one of the things, that I remember from back in the 228 00:10:47,070 --> 00:10:51,540 day, like I, I don't remember really thinking about this lately, but back 229 00:10:51,540 --> 00:10:56,205 in the day, I would say that for every gigabyte of primary storage, you 230 00:10:56,205 --> 00:10:58,695 had 20 gigabytes of backup storage. 231 00:10:59,255 --> 00:10:59,535 Mm-hmm. 232 00:10:59,535 --> 00:11:06,105 And so if you're gonna do that with disk, even, even once, many years ago. 233 00:11:06,225 --> 00:11:06,645 Wow. 234 00:11:06,645 --> 00:11:10,205 At this point, it's 20 years ago, but even once they came 235 00:11:10,225 --> 00:11:13,335 out with this idea of, SATA disk 236 00:11:13,605 --> 00:11:13,965 Yep. 237 00:11:14,415 --> 00:11:14,625 of 238 00:11:14,685 --> 00:11:15,075 nearline 239 00:11:15,075 --> 00:11:15,855 storage. 240 00:11:16,115 --> 00:11:18,665 that, that helped bring the cost down significantly. 241 00:11:19,095 --> 00:11:20,595 But not as much as deduplication. 242 00:11:20,880 --> 00:11:21,210 Yeah. 243 00:11:21,240 --> 00:11:23,850 Because even with those price differences, right? 244 00:11:23,850 --> 00:11:27,690 Maybe it was half the price or a third of the price, but once you add in that 245 00:11:27,690 --> 00:11:29,550 20 x that you talked about, Curtis, 246 00:11:29,565 --> 00:11:30,015 Yeah. 247 00:11:30,240 --> 00:11:31,050 that adds up. 248 00:11:31,440 --> 00:11:33,870 And it's not only just the storage cost, it's also you have 249 00:11:33,870 --> 00:11:37,950 to account for the power, the cooling, the floor space, right? 250 00:11:37,950 --> 00:11:39,620 All the things that go into that system. 251 00:11:40,830 --> 00:11:41,280 Yeah. 252 00:11:41,460 --> 00:11:41,880 Yeah. 253 00:11:42,390 --> 00:11:45,440 it's funny, just a, an afterthought that, that. 254 00:11:46,550 --> 00:11:50,910 Post that, that Chris Mellor did about the 30 years. 255 00:11:51,360 --> 00:11:55,230 The one group that jumped on the article and just started retweeting 256 00:11:55,230 --> 00:12:00,530 all kinds of parts of, or pieces of the article was the tape group , because 257 00:12:00,580 --> 00:12:02,440 I said really good things about tape. 258 00:12:02,810 --> 00:12:08,125 and the thing is that, I, I believe in all of those things, but. 259 00:12:09,845 --> 00:12:14,285 all of the advancements that I've seen in backup in the last 20 plus 260 00:12:14,285 --> 00:12:16,955 years has been disk and deduplication. 261 00:12:17,725 --> 00:12:22,765 so let's talk about, so what, so not everybody really 262 00:12:22,770 --> 00:12:24,385 understands what deduplication is. 263 00:12:24,655 --> 00:12:30,025 Some people used to describe it like, it's like compression, 264 00:12:30,335 --> 00:12:33,345 the way I remember it's like macro compression. 265 00:12:34,065 --> 00:12:34,965 it's like compression over 266 00:12:34,995 --> 00:12:35,235 Yeah. 267 00:12:36,165 --> 00:12:36,885 What do you think of that? 268 00:12:38,865 --> 00:12:40,005 I don't quite like that, 269 00:12:40,290 --> 00:12:40,680 having, 270 00:12:40,710 --> 00:12:44,310 some old blog posts that I might have said that phrase, but go ahead. 271 00:12:44,730 --> 00:12:46,695 so in my mind, deduplication is. 272 00:12:47,890 --> 00:12:54,430 Finding two identical segments and tossing one away, keeping only one copy, but still 273 00:12:54,430 --> 00:12:58,960 keeping a reference to that so you can, so you still know you have two virtual 274 00:12:58,965 --> 00:13:00,980 copies, but one physical copy, right? 275 00:13:01,570 --> 00:13:03,160 At a high level, that's what I, and now 276 00:13:03,330 --> 00:13:03,450 you? 277 00:13:04,750 --> 00:13:11,410 what is compression is taking an object, a singular object, and 278 00:13:11,410 --> 00:13:13,030 squeezing it into a smaller space. 279 00:13:14,120 --> 00:13:16,640 But how do you understand how compression works? 280 00:13:16,640 --> 00:13:17,940 Cuz I Sure as hell don't 281 00:13:18,290 --> 00:13:21,200 yeah, so typically like you would run it through different types of algorithms 282 00:13:21,200 --> 00:13:26,420 like LZ compression and all the rest in order to look for patterns and 283 00:13:26,420 --> 00:13:29,450 throw away bits and compress it down. 284 00:13:29,450 --> 00:13:31,640 Now, the difference I would say between duping compression 285 00:13:31,640 --> 00:13:32,810 because they do sound the same, 286 00:13:33,300 --> 00:13:33,590 Yeah. 287 00:13:33,610 --> 00:13:33,960 right? 288 00:13:34,640 --> 00:13:39,630 I would say one of the differences is with deduplication. 289 00:13:42,035 --> 00:13:45,335 It's more like a file system level compression, if you want to 290 00:13:45,335 --> 00:13:48,035 think of it that way, because it's not just I'm taking this block. 291 00:13:48,815 --> 00:13:49,115 Yeah. 292 00:13:49,325 --> 00:13:52,085 It's not just I'm taking this and I'm squeezing it down such 293 00:13:52,090 --> 00:13:54,305 that it could be, I just need to look at this and figure it out. 294 00:13:54,305 --> 00:13:57,035 It's a lot more complex than that. 295 00:13:57,155 --> 00:13:59,115 It is definitely a lot more complex than compression. 296 00:13:59,625 --> 00:14:04,155 I've just, I've just honestly never dug into the code of how 297 00:14:04,635 --> 00:14:06,225 traditional compression works. 298 00:14:06,625 --> 00:14:11,605 So the idea is that I'm looking for duplicate segments of data across many 299 00:14:11,605 --> 00:14:16,855 places, both from different sources as well as different time periods, right? 300 00:14:16,885 --> 00:14:17,200 Yep. 301 00:14:18,295 --> 00:14:22,305 the, this chunk of data that's coming in right now and tonight's backup. 302 00:14:22,365 --> 00:14:27,255 I'm comparing it literally with every chunk of data that I've ever received 303 00:14:28,350 --> 00:14:28,650 It. 304 00:14:28,845 --> 00:14:29,175 else. 305 00:14:29,535 --> 00:14:32,505 . I would say that's an ideal system, but not everyone builds 306 00:14:32,505 --> 00:14:33,735 their deduplication that way. 307 00:14:34,530 --> 00:14:35,160 where 308 00:14:35,430 --> 00:14:35,670 it's 309 00:14:35,790 --> 00:14:36,090 in. 310 00:14:36,764 --> 00:14:37,575 yeah, go ahead. 311 00:14:38,100 --> 00:14:38,310 Yeah. 312 00:14:38,310 --> 00:14:42,175 So it all goes down to what is your deduplication domain is another term 313 00:14:42,175 --> 00:14:43,945 that some people talk about, right? 314 00:14:43,945 --> 00:14:45,985 Which is it limited to a system? 315 00:14:45,985 --> 00:14:48,955 Is it limited to a cluster which might be formed to multiple 316 00:14:48,955 --> 00:14:53,574 systems, or is it limited to a single backup stream coming in? 317 00:14:54,265 --> 00:14:54,625 Right. 318 00:14:54,625 --> 00:14:54,955 So 319 00:14:54,955 --> 00:14:55,225 there. 320 00:14:55,359 --> 00:14:57,579 that the question is what is your data domain? 321 00:14:57,669 --> 00:14:57,999 Yeah. 322 00:14:59,049 --> 00:15:00,039 D Domain. 323 00:15:00,709 --> 00:15:01,669 So let's back up. 324 00:15:02,719 --> 00:15:06,679 So a, as I understand it, right, so basically we're taking the data 325 00:15:06,679 --> 00:15:10,159 that's, that's coming in or that's going to come in, we're slicing 326 00:15:10,159 --> 00:15:11,909 it up into, I like the term chunk. 327 00:15:12,754 --> 00:15:13,099 , right? 328 00:15:13,579 --> 00:15:17,449 We run those chunks through a cryptographic hashing algorithm. 329 00:15:18,109 --> 00:15:21,709 one, Shaw 2 56, whatever it, whatever you're using. 330 00:15:22,099 --> 00:15:27,809 On the other side of that, we get a alpha numeric value, in the case of SH one, 331 00:15:27,814 --> 00:15:31,049 it's 160 bit alpha alphanumeric value. 332 00:15:31,589 --> 00:15:37,020 so basically you, depending on the algorithm you use, you get a, you get an 333 00:15:37,020 --> 00:15:42,330 alpha numeric value at the end, and the size of that val, of that value is going 334 00:15:42,330 --> 00:15:44,939 to be based on which algorithm you use. 335 00:15:44,939 --> 00:15:48,719 In the case of SHA-1, 160 bits, right? 336 00:15:50,370 --> 00:15:53,130 You can then take the 160 bits. 337 00:15:53,130 --> 00:15:54,720 You can't reverse engineer it. 338 00:15:54,720 --> 00:15:59,550 You can't take the 160 bits and turn it into the chunk, but you can use that 339 00:15:59,550 --> 00:16:02,700 value to uniquely identify that chunk. 340 00:16:03,265 --> 00:16:03,485 Yep. 341 00:16:03,515 --> 00:16:09,600 have another chunk of data, regardless of where it came from, If it's 160 342 00:16:09,600 --> 00:16:11,760 bit value, again, that's SHA-1 and 343 00:16:12,470 --> 00:16:12,635 Yep. 344 00:16:12,870 --> 00:16:13,140 are different. 345 00:16:13,350 --> 00:16:19,410 If it's fingerprint is the same, you can say that this chunk is identical to that 346 00:16:19,485 --> 00:16:19,815 Yep. 347 00:16:20,430 --> 00:16:25,270 that had the same fingerprint, and you can then discard other chunk, right? 348 00:16:25,800 --> 00:16:26,099 Yeah, 349 00:16:26,199 --> 00:16:30,099 you can discard the actual data, but you should still keep track 350 00:16:30,105 --> 00:16:33,429 of it somewhere in a file system, just because you need, still need 351 00:16:33,479 --> 00:16:33,659 Yeah. 352 00:16:33,659 --> 00:16:34,499 You're gonna keep track. 353 00:16:34,679 --> 00:16:36,299 Oh, we found another one of these, 354 00:16:36,540 --> 00:16:36,750 Yeah. 355 00:16:37,950 --> 00:16:41,700 And so usually that lookup is in a deduplication index 356 00:16:41,700 --> 00:16:42,450 is what they called them. 357 00:16:42,450 --> 00:16:43,829 Usually a dedupe index, Right. 358 00:16:43,890 --> 00:16:47,095 which keeps a list of, Hey, here are all the fingerprints that I have. 359 00:16:47,635 --> 00:16:52,494 As we, we were alluding to before, one of the things that determines 360 00:16:52,494 --> 00:16:57,155 sort of your effectiveness of dedupe is the dedupe domain, right? 361 00:16:57,155 --> 00:17:00,935 So I've seen it file system level, meaning it only looks for 362 00:17:00,940 --> 00:17:02,645 duplicate data within each volume. 363 00:17:03,545 --> 00:17:07,115 I've seen it host level, I've seen it backup level, meaning 364 00:17:07,115 --> 00:17:09,125 literally backup configuration wise. 365 00:17:10,055 --> 00:17:10,235 Yeah. 366 00:17:10,325 --> 00:17:13,864 So if I have a Windows server and I'm backing up the host and I'm 367 00:17:13,864 --> 00:17:18,484 backing up SQL Server, I only look for duplicates within SQL Server 368 00:17:18,514 --> 00:17:20,704 backups right against each other. 369 00:17:21,314 --> 00:17:27,065 then we have, if we're backing up several systems to a box, right? 370 00:17:27,515 --> 00:17:27,865 Maybe 371 00:17:27,875 --> 00:17:28,295 Mm-hmm. 372 00:17:28,554 --> 00:17:30,705 the dedupe domain is only within that box. 373 00:17:30,705 --> 00:17:31,905 It's only looking for. 374 00:17:33,074 --> 00:17:34,604 Duplicates between all of that. 375 00:17:34,844 --> 00:17:39,894 And then there's what I would call truly global dedupe, which is we're 376 00:17:39,899 --> 00:17:45,455 looking for duplicates from everything coming in, from multiple sources. 377 00:17:45,554 --> 00:17:52,769 there is Point of decreasing marginal returns, right? 378 00:17:52,860 --> 00:17:57,010 You can argue, and certainly if you're a company that only does d dedupe within, 379 00:17:57,030 --> 00:18:00,330 like earlier I was, we only looked for dupes within SQL server backups. 380 00:18:00,485 --> 00:18:00,705 yep. 381 00:18:00,959 --> 00:18:04,165 You could make an argument that, there's not a lot of duplicate data 382 00:18:04,169 --> 00:18:05,394 between SQL Server and Windows, 383 00:18:06,295 --> 00:18:06,730 Yep. 384 00:18:06,864 --> 00:18:07,084 so 385 00:18:07,480 --> 00:18:08,830 And so you keep 'em separate. 386 00:18:08,844 --> 00:18:09,384 two, 387 00:18:09,725 --> 00:18:11,735 not gonna be a lot of duplicate data there, and there's not 388 00:18:11,735 --> 00:18:12,784 gonna be a lot of duplicate. 389 00:18:13,944 --> 00:18:17,659 between the SQL Server database on this host and the SQL 390 00:18:17,659 --> 00:18:19,189 Server database on that host. 391 00:18:19,325 --> 00:18:19,565 Yeah, 392 00:18:19,699 --> 00:18:21,079 that's another argument that some 393 00:18:21,079 --> 00:18:21,345 make 394 00:18:21,395 --> 00:18:25,475 but I think a lot of that was because of architectural limitations of the 395 00:18:25,480 --> 00:18:29,735 products themselves rather than, that is really what you wanted to do. 396 00:18:29,735 --> 00:18:30,035 Because 397 00:18:30,035 --> 00:18:31,235 that's more of a management issue. 398 00:18:31,969 --> 00:18:35,530 they didn't, it was like, if we're gonna do it, if we're gonna do it 399 00:18:35,679 --> 00:18:38,379 that way, it's gonna be much harder. 400 00:18:38,845 --> 00:18:39,135 Yeah. 401 00:18:39,145 --> 00:18:41,365 to design a product to do it that way. 402 00:18:42,175 --> 00:18:44,485 And we don't think, we don't think that there's going to 403 00:18:44,485 --> 00:18:45,865 be that much more benefit, 404 00:18:46,465 --> 00:18:49,795 But on the other hand, if you look at things like VMware, right? 405 00:18:49,795 --> 00:18:52,575 If I have a bunch of VMs, there's a good cha, and they all came 406 00:18:52,575 --> 00:18:54,255 from a single golden image, right? 407 00:18:54,255 --> 00:18:58,545 There's a good chance that as you're backing it up, 80, 90% of that stuff 408 00:18:58,545 --> 00:19:00,225 is all gonna be deduplicated, right? 409 00:19:00,869 --> 00:19:01,709 Absolutely. 410 00:19:01,919 --> 00:19:02,369 Yeah. 411 00:19:02,699 --> 00:19:07,985 There's also a lot of duplicate data even within a large filer, right? 412 00:19:07,989 --> 00:19:08,374 There's gonna 413 00:19:08,435 --> 00:19:08,675 Yeah. 414 00:19:08,854 --> 00:19:10,054 of duplicate data there, right? 415 00:19:10,054 --> 00:19:13,745 So if you're only doing it volume to volume or backup configuration 416 00:19:13,745 --> 00:19:16,445 to backup configuration, you, there's a lot of duplicate data 417 00:19:16,445 --> 00:19:17,324 that I think you would miss. 418 00:19:17,415 --> 00:19:17,685 Yeah. 419 00:19:18,314 --> 00:19:21,225 I know you talked about the domains, but I think another thing to also 420 00:19:21,225 --> 00:19:26,204 mention is, Some products do different types of chunking, if you will. 421 00:19:26,384 --> 00:19:31,624 Some do it at the file level, others do it at a smaller level, right? 422 00:19:31,629 --> 00:19:35,004 And some do fixed segment where each one is a fixed length. 423 00:19:35,334 --> 00:19:38,994 Others do variable segments where they try to figure out what is optimal, 424 00:19:38,999 --> 00:19:42,984 because depending on how you're doing your fingerprinting, you want to find 425 00:19:42,984 --> 00:19:44,994 the most number of matches, right? 426 00:19:44,994 --> 00:19:46,464 So you can save on storage. 427 00:19:47,199 --> 00:19:48,099 another thing that also comes up. 428 00:19:48,984 --> 00:19:51,904 I would argue that file level dedupe isn't really dedupe, 429 00:19:51,929 --> 00:19:53,514 it's more a single instance. 430 00:19:53,934 --> 00:19:56,244 that's like single instance storage of a file, 431 00:19:56,604 --> 00:19:57,114 Okay. 432 00:19:57,274 --> 00:19:57,934 it's, yeah. 433 00:19:58,234 --> 00:20:01,584 But I'm always thinking subfile, when 434 00:20:01,654 --> 00:20:01,874 Yep. 435 00:20:02,634 --> 00:20:10,434 what I think of actual dedupe . There is a much, like a very big, other 436 00:20:10,454 --> 00:20:16,354 way that we divide up the dedupe industry, that is source versus target. 437 00:20:16,864 --> 00:20:17,084 Yep. 438 00:20:18,499 --> 00:20:23,689 the, the first dedupe product I ever saw, 439 00:20:23,779 --> 00:20:24,479 Hydro store. 440 00:20:24,564 --> 00:20:26,949 was, no, was not, 441 00:20:27,119 --> 00:20:27,339 No. 442 00:20:27,459 --> 00:20:32,079 not the first, no, the first one I saw the product at the time was called Undo. 443 00:20:32,849 --> 00:20:33,609 Have we talked 444 00:20:33,609 --> 00:20:33,929 Interesting. 445 00:20:34,389 --> 00:20:35,619 Undo with two Os. 446 00:20:35,649 --> 00:20:39,579 It was really funny that the name of a dedupe vendor. 447 00:20:40,914 --> 00:20:43,644 Had duplicate data in their company name. 448 00:20:44,124 --> 00:20:45,894 It was undoo with two os. 449 00:20:46,734 --> 00:20:49,849 You know this product, you just don't know that's what it used to be called. 450 00:20:50,404 --> 00:20:50,764 What is it? 451 00:20:52,954 --> 00:20:53,134 What 452 00:20:53,134 --> 00:20:53,799 is the product? 453 00:20:53,849 --> 00:20:55,439 I'll give you a hint. 454 00:20:56,429 --> 00:20:56,759 It. 455 00:20:57,719 --> 00:21:03,269 The name comes from the fact that it would be a C of availability. 456 00:21:06,774 --> 00:21:08,704 I'm gonna, I'm gonna put the Jeopardy theme in here. 457 00:21:11,389 --> 00:21:15,019 What would it see of availability? 458 00:21:15,639 --> 00:21:18,124 That's what the name, that's where the name for the company comes 459 00:21:18,124 --> 00:21:23,884 from, or if I want to put it in the right order, an availability c. 460 00:21:25,574 --> 00:21:27,044 I don't know what this is. 461 00:21:28,424 --> 00:21:28,844 Avamar 462 00:21:29,614 --> 00:21:30,764 oh, that makes sense. 463 00:21:31,694 --> 00:21:32,054 Yeah. 464 00:21:32,804 --> 00:21:34,239 that's where the name Avamar came from. 465 00:21:34,599 --> 00:21:35,529 So the first 466 00:21:35,799 --> 00:21:36,549 I should know that 467 00:21:37,709 --> 00:21:38,529 you shouldn't know 468 00:21:38,634 --> 00:21:42,274 I having being, part of my former employer. 469 00:21:42,484 --> 00:21:42,844 Yes. 470 00:21:42,859 --> 00:21:43,369 Yeah. 471 00:21:43,499 --> 00:21:47,089 I, I have a bit of an inside track because that they're, They 472 00:21:47,089 --> 00:21:48,679 were right up the road from me, 473 00:21:48,844 --> 00:21:49,064 Yep. 474 00:21:49,159 --> 00:21:49,759 were up there. 475 00:21:49,789 --> 00:21:50,629 They were up in Irvine. 476 00:21:51,349 --> 00:21:53,319 and that was, the first dedupe product. 477 00:21:53,349 --> 00:21:57,839 They were a source dedupe . So what's the difference between source 478 00:21:57,839 --> 00:21:59,649 dedupe and target dedupe Prasanna? 479 00:22:00,184 --> 00:22:04,774 So the biggest one is, so let's first talk about target tup, right? 480 00:22:04,774 --> 00:22:10,024 So Target Tup is data comes into the system and then a deduplication 481 00:22:10,024 --> 00:22:12,094 algorithm runs tosses away data. 482 00:22:12,494 --> 00:22:15,314 It can support any type of client as long as it supports 483 00:22:15,319 --> 00:22:16,394 whatever the protocol it has. 484 00:22:16,414 --> 00:22:19,484 So it's NFS or smb, right? 485 00:22:19,844 --> 00:22:21,844 Whatever can write to it, the data gets deduped. 486 00:22:22,074 --> 00:22:22,314 hang on. 487 00:22:22,314 --> 00:22:23,304 Before you go on to that. 488 00:22:23,564 --> 00:22:25,004 I don't disagree with what you said. 489 00:22:25,004 --> 00:22:27,404 I just, I think there could be a little bit more clarification. 490 00:22:27,794 --> 00:22:31,485 It's a box I send whatever I want to. 491 00:22:32,434 --> 00:22:32,654 Yep. 492 00:22:32,939 --> 00:22:37,769 Typically it, the thing about Target Dedup was that, that it was, you didn't 493 00:22:37,769 --> 00:22:39,239 have to do a lot of re-engineering of 494 00:22:39,324 --> 00:22:39,404 seamless. 495 00:22:39,764 --> 00:22:40,064 Yeah, 496 00:22:40,499 --> 00:22:40,859 Right. 497 00:22:40,934 --> 00:22:42,194 a VTL system, right? 498 00:22:42,194 --> 00:22:42,404 That came. 499 00:22:42,509 --> 00:22:43,559 plug in a box. 500 00:22:43,619 --> 00:22:44,099 Yeah. 501 00:22:44,129 --> 00:22:48,419 And you would send you, and basically you stopped using tape and you 502 00:22:48,419 --> 00:22:49,509 sent your backups to this box. 503 00:22:49,799 --> 00:22:52,739 Maybe the box might even be pretending to be a tape library, 504 00:22:52,744 --> 00:22:53,999 the virtual tape library. 505 00:22:54,049 --> 00:22:56,569 and then it did all the dedupe magic over there. 506 00:22:57,139 --> 00:22:59,969 Which was great because you can just plug in your box and go. 507 00:23:00,614 --> 00:23:01,034 Yeah. 508 00:23:01,424 --> 00:23:04,634 the other side is called source side dedupe, instead of sending all the data 509 00:23:04,634 --> 00:23:07,964 and tossing it away, why don't we do something smart and actually figure out 510 00:23:07,964 --> 00:23:13,364 the duplicates on the client itself, on the source right, dedupe on the 511 00:23:13,369 --> 00:23:15,114 source, and only send the unique data. 512 00:23:15,494 --> 00:23:17,864 And this has the advantage. 513 00:23:19,049 --> 00:23:21,989 Actually not sending the data over the wire, which is actually 514 00:23:21,989 --> 00:23:25,739 a huge benefit that people don't understand always, right? 515 00:23:25,739 --> 00:23:29,579 Is not sending the data can actually make it a lot faster, even though 516 00:23:29,579 --> 00:23:32,949 you think, oh, I'm now putting additional load on my server itself. 517 00:23:32,949 --> 00:23:38,229 But it ends up being better than trying to send all the data and just tossing 518 00:23:38,229 --> 00:23:40,089 it away like target-side dedupe does. 519 00:23:40,584 --> 00:23:42,924 I would say it theoretically should be better 520 00:23:44,004 --> 00:23:44,394 right? 521 00:23:44,634 --> 00:23:44,994 Because 522 00:23:45,039 --> 00:23:45,159 I. 523 00:23:45,894 --> 00:23:47,944 I'm just saying I've seen some crappy source 524 00:23:48,024 --> 00:23:48,324 Yeah. 525 00:23:48,774 --> 00:23:49,074 right? 526 00:23:49,134 --> 00:23:49,434 Okay. 527 00:23:49,644 --> 00:23:49,854 Sorry. 528 00:23:49,854 --> 00:23:52,614 I've seen some, I've seen some good ones, or the ones that I've 529 00:23:52,614 --> 00:23:53,994 interacted with have been good. 530 00:23:53,994 --> 00:23:56,964 And so I've seen the performance numbers around 531 00:23:57,414 --> 00:23:57,834 turning it 532 00:23:57,929 --> 00:24:00,539 I do think it, it makes more sense to me. 533 00:24:00,659 --> 00:24:02,309 It always made more sense to me. 534 00:24:02,519 --> 00:24:07,449 The only reason why we had Target dedupe was because to do source dedupe , you 535 00:24:07,449 --> 00:24:08,669 have to redesign the backup product. 536 00:24:09,744 --> 00:24:10,034 , Yeah. 537 00:24:10,139 --> 00:24:15,229 It took a long time to get, basically you have to stop using net backup networker 538 00:24:15,414 --> 00:24:15,654 Yeah. 539 00:24:16,089 --> 00:24:19,029 whatever it was back in the day, and you had to replace it. 540 00:24:19,059 --> 00:24:20,829 Like in this case with Avamar, Avamar 541 00:24:20,934 --> 00:24:21,264 Yep. 542 00:24:22,659 --> 00:24:23,379 do-do product. 543 00:24:23,379 --> 00:24:25,179 You had to do what we call a four clipped upgrade. 544 00:24:25,179 --> 00:24:26,079 You had to throw out the baby with 545 00:24:26,124 --> 00:24:26,304 Yep. 546 00:24:26,649 --> 00:24:27,349 whatever phrase, whatever. 547 00:24:28,674 --> 00:24:30,144 analogy you want to use there. 548 00:24:30,594 --> 00:24:34,104 That was the main problem as I saw it with source dedup. 549 00:24:34,154 --> 00:24:38,319 is that you had to change your backup product to get it, 550 00:24:38,529 --> 00:24:40,959 Well, and that was in the beginning, right? 551 00:24:41,439 --> 00:24:42,339 At the very early 552 00:24:42,409 --> 00:24:42,709 You. 553 00:24:42,709 --> 00:24:43,144 You, yeah. 554 00:24:43,144 --> 00:24:46,014 Now you just had to upgrade your backup product, 555 00:24:46,034 --> 00:24:46,324 Yeah. 556 00:24:46,404 --> 00:24:54,204 Because many of modern backup technologies now support source dedupe , although even 557 00:24:54,204 --> 00:24:54,624 all. 558 00:24:54,989 --> 00:24:57,894 backup technologies don't, I don't 559 00:24:57,894 --> 00:24:57,984 Yeah. 560 00:24:58,164 --> 00:25:01,674 I dunno if that came out in English, so some I, there was 561 00:25:01,674 --> 00:25:02,934 some double negatives in there. 562 00:25:03,594 --> 00:25:08,934 Some very new backup technologies. 563 00:25:09,534 --> 00:25:11,344 Don't do source dedupe 564 00:25:11,700 --> 00:25:12,870 . which seems bunkers. 565 00:25:13,070 --> 00:25:14,630 which does seem bonkers. 566 00:25:15,180 --> 00:25:19,210 and, I'm talking about the likes of Rubric and Cohesity, right? 567 00:25:19,215 --> 00:25:24,230 These are new, these are, next gen backup products that were designed 568 00:25:24,230 --> 00:25:27,380 in the last, than the last 10 years. 569 00:25:27,950 --> 00:25:30,020 And it's based on an appliance model. 570 00:25:30,760 --> 00:25:35,515 and they do all the dedupe inside that box, is my understanding, right? 571 00:25:36,785 --> 00:25:41,105 And I just wanna challenge that, Curtis, because I thought in some cases, 572 00:25:41,395 --> 00:25:41,815 Mm-hmm. 573 00:25:41,825 --> 00:25:47,345 They do source side deduplication, but I think because they've tried to 574 00:25:47,345 --> 00:25:51,130 be open and act as a target device, in those cases, you can't, you 575 00:25:51,130 --> 00:25:52,390 don't really have another option. 576 00:25:53,155 --> 00:25:55,585 Yeah, I don't, again, I'm not, 577 00:25:55,600 --> 00:25:56,230 I, but I don't know 578 00:25:56,420 --> 00:25:57,890 work at Druva, not at Rubrik, 579 00:25:58,170 --> 00:25:59,215 or Cohesity. 580 00:25:59,215 --> 00:26:02,815 But it is my understanding that they do target side dedup, which is, 581 00:26:03,075 --> 00:26:07,325 and one of the challenges of target side dedup is need an appliance. 582 00:26:08,515 --> 00:26:09,870 at each location. 583 00:26:10,350 --> 00:26:13,020 Now I know that they can do virtual appliances, right? 584 00:26:13,020 --> 00:26:14,790 So they have a VM level appliance. 585 00:26:15,190 --> 00:26:20,350 but you need a box or something pretending to be a box at each location, because 586 00:26:20,350 --> 00:26:26,560 if you're not eliminating the duplicates before you send it to the box, then you 587 00:26:26,820 --> 00:26:28,800 need something that's on-prem, right? 588 00:26:29,295 --> 00:26:31,235 Because you definitely don't wanna send that all over the Wan 589 00:26:31,860 --> 00:26:35,550 no, that's the, to me, that's the biggest advantage of a source dedupe 590 00:26:35,570 --> 00:26:38,030 system is that it's ultimately scalable, 591 00:26:38,455 --> 00:26:38,675 Yep. 592 00:26:38,780 --> 00:26:42,620 That you, that assuming it doesn't slow things down, all these things, 593 00:26:42,620 --> 00:26:48,270 assuming that the product actually works, that you, you could back up a laptop. 594 00:26:49,000 --> 00:26:49,365 , right? 595 00:26:49,365 --> 00:26:53,550 You can back up a mobile phone and the duplicate data will be eliminated 596 00:26:53,550 --> 00:26:56,520 before it's sent over the wan, which is what you need to do if you're 597 00:26:56,525 --> 00:26:57,990 backing up something over the internet. 598 00:26:59,210 --> 00:27:05,165 so the downside that some, again, you talked about it already, is that it does 599 00:27:05,165 --> 00:27:09,905 put additional compute on the client. 600 00:27:12,155 --> 00:27:15,275 The argument is that it's offset by the, 601 00:27:15,725 --> 00:27:17,735 the savings of the network bandwidth. 602 00:27:17,785 --> 00:27:19,045 There is also one more downside, 603 00:27:19,795 --> 00:27:19,945 okay. 604 00:27:20,125 --> 00:27:21,175 which is that. 605 00:27:22,075 --> 00:27:26,365 Not all applications can do source side deduplication. 606 00:27:26,935 --> 00:27:32,875 So if you do have an application which only supports writing to like an NFS Mount 607 00:27:32,875 --> 00:27:38,125 point or an SMB Mount point, or something that doesn't allow the integration of 608 00:27:38,125 --> 00:27:41,590 these source side duplication logic, then you are going to need to be 609 00:27:41,590 --> 00:27:42,950 able to support target side dedupe. 610 00:27:42,950 --> 00:27:43,000 do. 611 00:27:43,870 --> 00:27:44,320 Yep. 612 00:27:44,780 --> 00:27:45,320 agreed. 613 00:27:45,710 --> 00:27:49,425 and an example of that would be like, Oracle, right? 614 00:27:49,545 --> 00:27:49,905 Yep. 615 00:27:50,355 --> 00:27:50,715 Right. 616 00:27:51,015 --> 00:27:51,645 merge. 617 00:27:52,875 --> 00:27:53,295 yeah. 618 00:27:54,015 --> 00:27:57,145 I would think that you should be able, I don't know, we 619 00:27:57,205 --> 00:27:57,905 No, you can't. 620 00:27:57,905 --> 00:27:58,345 You can't. 621 00:27:58,445 --> 00:27:58,905 You can't. 622 00:27:59,365 --> 00:28:02,755 You can't take the Oracle stream and slice it and dice it. 623 00:28:06,625 --> 00:28:07,225 I don't know. 624 00:28:07,240 --> 00:28:07,900 Did you what? 625 00:28:08,110 --> 00:28:08,410 Sorry? 626 00:28:08,560 --> 00:28:12,850 You could, there are companies out there which give, which provide 627 00:28:12,850 --> 00:28:14,770 a virtual file system interface 628 00:28:15,460 --> 00:28:16,120 that lives 629 00:28:16,120 --> 00:28:16,705 on the client. 630 00:28:16,810 --> 00:28:16,990 it. 631 00:28:16,990 --> 00:28:17,410 You fake 632 00:28:17,485 --> 00:28:17,755 Yeah. 633 00:28:18,190 --> 00:28:18,430 Yeah. 634 00:28:18,565 --> 00:28:18,805 Yep. 635 00:28:20,050 --> 00:28:20,440 All right. 636 00:28:20,440 --> 00:28:23,560 And then I've got something called hybrid dedupe and this was 637 00:28:23,565 --> 00:28:25,270 invented by your former employer. 638 00:28:27,225 --> 00:28:29,115 I don't even know what a hybrid dedupe is. 639 00:28:30,240 --> 00:28:32,310 it's Target Dedoo pretending to be Source cdu. 640 00:28:35,685 --> 00:28:45,375 Target D. Oh, see, here's my, okay, so here's my problem is I think Boost 641 00:28:45,840 --> 00:28:46,190 Uhhuh. 642 00:28:47,025 --> 00:28:50,295 is source. 643 00:28:50,295 --> 00:28:55,035 I deduplication, I don't know if I would call it hybrid, because it is 644 00:28:55,040 --> 00:28:57,075 very similar to what Avamar DI did. 645 00:28:58,355 --> 00:28:58,650 , right? 646 00:28:58,655 --> 00:29:02,190 It's moving the deduplication logic to the client 647 00:29:02,495 --> 00:29:02,915 Mm-hmm. 648 00:29:03,150 --> 00:29:05,640 such that you could do all of the computation. 649 00:29:05,640 --> 00:29:08,910 The same thing that we have talked about with source I deduplication, 650 00:29:09,675 --> 00:29:11,385 I'll tell you why I put it in a different category. 651 00:29:11,965 --> 00:29:15,255 To me, hybrid dedupe is redoing the backup software. 652 00:29:16,095 --> 00:29:19,485 I'm sorry, source dedupe, source iDation. 653 00:29:20,595 --> 00:29:21,255 It's done at 654 00:29:21,260 --> 00:29:21,550 Okay. 655 00:29:21,675 --> 00:29:22,455 software level, 656 00:29:22,950 --> 00:29:23,760 Okay, then. 657 00:29:23,760 --> 00:29:23,850 I 658 00:29:23,905 --> 00:29:30,015 with hybrid dedupe . I'm still dumb sending everything to this source dedupe 659 00:29:30,075 --> 00:29:31,885 thing that's gonna redo it, right? 660 00:29:32,165 --> 00:29:33,630 doesn't matter in the end, you get 661 00:29:33,720 --> 00:29:34,410 the benefits. 662 00:29:34,920 --> 00:29:35,820 benefits, right? 663 00:29:36,280 --> 00:29:37,310 that's what, 664 00:29:37,420 --> 00:29:37,750 Okay. 665 00:29:37,750 --> 00:29:38,620 So with hybrid, yeah. 666 00:29:38,620 --> 00:29:42,940 You get the benefits of source without having to upgrade and, or sorry, 667 00:29:42,945 --> 00:29:44,260 throw away your backup software. 668 00:29:44,860 --> 00:29:49,740 so I, we spent most of this time talking about dedupe there are 669 00:29:49,745 --> 00:29:52,410 a bunch of different ways to use disk in your backup system. 670 00:29:53,460 --> 00:29:55,770 Some of which don't really require dedup, right? 671 00:29:55,770 --> 00:29:58,860 We used to do what we call disk cashing, where you just had enough 672 00:29:58,860 --> 00:30:00,330 disk for last night's backup. 673 00:30:00,330 --> 00:30:03,840 You would back up to disk and then you would copy that to tape, and then 674 00:30:03,840 --> 00:30:05,340 you would hand that to a man in a van. 675 00:30:05,860 --> 00:30:07,030 we got a bunch of different things. 676 00:30:07,030 --> 00:30:10,565 I got D to T D to D to C, and D to C. Did I do all that? 677 00:30:10,925 --> 00:30:14,775 So dis to tape disc, to disc to disk, direct cloud and 678 00:30:14,780 --> 00:30:15,885 dis to disc to cloud, right? 679 00:30:15,885 --> 00:30:21,645 So these are all that people use disk in current backup systems. 680 00:30:22,155 --> 00:30:25,861 to me, D to C or disto disc to cloud is really dis to disc. 681 00:30:26,040 --> 00:30:27,720 To disc is just the cloud is or the 682 00:30:27,840 --> 00:30:28,020 Yeah. 683 00:30:28,020 --> 00:30:28,740 Is a desk. 684 00:30:28,920 --> 00:30:29,220 Yeah. 685 00:30:29,700 --> 00:30:30,120 right? 686 00:30:30,510 --> 00:30:36,675 And I will say that dedupe , by the way, I will say that without d. The 687 00:30:36,675 --> 00:30:41,175 whole thing of using the cloud, the way we use the cloud just wouldn't work. 688 00:30:42,445 --> 00:30:44,275 you can't send full backups to the cloud. 689 00:30:44,280 --> 00:30:45,025 you could 690 00:30:45,130 --> 00:30:50,050 you could, but it would be well, and yeah, with unlimited bandwidth 691 00:30:50,055 --> 00:30:51,370 it would just be expensive. 692 00:30:51,370 --> 00:30:53,620 Just going back to the conversation we had earlier about the wan, right? 693 00:30:54,070 --> 00:30:57,520 You don't wanna send full copies out to over the wan. 694 00:30:58,235 --> 00:31:01,025 because that gets expensive and very slow. 695 00:31:01,395 --> 00:31:06,265 the other one I was going to comment on was, oh, I know we've been 696 00:31:06,270 --> 00:31:10,195 talking about disk, but I think it's also important to acknowledge that 697 00:31:10,195 --> 00:31:13,555 now it's no longer spinning disk. 698 00:31:13,555 --> 00:31:14,665 It could also be flash. 699 00:31:15,445 --> 00:31:16,305 We've seen 700 00:31:16,935 --> 00:31:17,475 yeah, 701 00:31:17,475 --> 00:31:17,955 but that's a 702 00:31:17,985 --> 00:31:18,465 I know. 703 00:31:18,465 --> 00:31:18,915 thing 704 00:31:19,065 --> 00:31:22,725 I, know, but I'm just saying that when it comes to deduplication and backup 705 00:31:22,730 --> 00:31:25,680 ST or protection storage, right? 706 00:31:26,100 --> 00:31:30,540 This, it could be flash, it could be disk, it could be object storage, right? 707 00:31:31,290 --> 00:31:34,290 So I think it's important to differentiate that, like what we're 708 00:31:34,290 --> 00:31:37,080 talking about with deduplication, when we mentioned disk, right? 709 00:31:37,080 --> 00:31:38,550 The media layer itself. 710 00:31:39,330 --> 00:31:40,620 Yeah, the media layer. 711 00:31:40,650 --> 00:31:41,070 Yes. 712 00:31:41,460 --> 00:31:42,961 The media layer is not tape. 713 00:31:44,630 --> 00:31:45,680 Hang on one second. 714 00:31:46,130 --> 00:31:49,565 I need to, didn't realize I had a, Meeting 715 00:31:49,655 --> 00:31:50,015 Meaning a. 716 00:31:51,755 --> 00:31:52,235 Yeah. 717 00:31:52,295 --> 00:31:52,715 Four. 718 00:31:52,715 --> 00:31:56,055 four 15, which is an odd, all right. 719 00:31:59,920 --> 00:32:02,500 it's a pre-meeting with a podcast thing. 720 00:32:02,500 --> 00:32:04,670 It's, anyway, yeah, okay, 721 00:32:06,600 --> 00:32:07,215 Just, just 722 00:32:07,410 --> 00:32:08,220 of flash 723 00:32:08,275 --> 00:32:09,245 flash and backup, 724 00:32:09,295 --> 00:32:11,395 I'm just saying that people will bring it up. 725 00:32:11,400 --> 00:32:14,215 So I just wanna clarify that when we talk about disc, we're 726 00:32:14,215 --> 00:32:15,385 just talking about not tape. 727 00:32:16,240 --> 00:32:17,920 The only place. 728 00:32:18,160 --> 00:32:19,030 Yeah. 729 00:32:19,060 --> 00:32:19,630 Correct. 730 00:32:19,960 --> 00:32:25,840 The only place where I think maybe Flash has a place in the backup 731 00:32:25,845 --> 00:32:31,120 system is, and the folks over at Pierre and Neil, mad at me now. 732 00:32:31,170 --> 00:32:36,710 But, the only place that I, where I think Flash has a place in the backup 733 00:32:36,710 --> 00:32:39,590 system is with like live recovery. 734 00:32:41,555 --> 00:32:42,125 If you're gonna 735 00:32:42,140 --> 00:32:42,440 Yep. 736 00:32:42,875 --> 00:32:46,265 if you're gonna do instant recovery and you're actually gonna run VMs 737 00:32:46,265 --> 00:32:50,855 off of your backups, that better be some really nice performing disk. 738 00:32:51,380 --> 00:32:51,620 Yeah. 739 00:32:51,755 --> 00:32:54,065 the thing is, it doesn't need to be your whole system. 740 00:32:54,065 --> 00:32:54,635 It just needs to 741 00:32:54,650 --> 00:32:55,040 Yeah. 742 00:32:55,520 --> 00:32:59,055 A part of, and it needs to, you don't need your entire system to be flash, 743 00:32:59,760 --> 00:33:00,050 Yeah. 744 00:33:00,075 --> 00:33:00,435 You just 745 00:33:00,440 --> 00:33:00,705 need 746 00:33:00,800 --> 00:33:01,090 yeah, 747 00:33:01,305 --> 00:33:03,016 enough to be able to support that use case. 748 00:33:03,660 --> 00:33:10,940 I just think that where Flash does really, Is in random access, right? 749 00:33:11,390 --> 00:33:13,760 Backup isn't a random access application. 750 00:33:15,050 --> 00:33:16,610 Backup is a streaming application. 751 00:33:16,610 --> 00:33:19,270 Even if what we're talking about large dedupe chunks. 752 00:33:19,490 --> 00:33:19,730 I don't 753 00:33:19,820 --> 00:33:20,090 yeah, 754 00:33:20,100 --> 00:33:20,220 I, 755 00:33:20,220 --> 00:33:22,800 say, let's just say the jury is out for me. 756 00:33:22,800 --> 00:33:23,820 I am in Missouri. 757 00:33:23,850 --> 00:33:24,420 Missouri. 758 00:33:24,720 --> 00:33:26,120 is that the show me state? 759 00:33:26,120 --> 00:33:26,990 That's the show me state. 760 00:33:27,040 --> 00:33:27,460 yeah. 761 00:33:28,660 --> 00:33:30,190 So I'll tell you what, I'll tell you what. 762 00:33:30,550 --> 00:33:34,540 If there's anybody that's listening to this that just got pissed off, 763 00:33:36,145 --> 00:33:36,535 what's his 764 00:33:36,535 --> 00:33:36,745 name? 765 00:33:36,745 --> 00:33:37,315 I'll come back 766 00:33:37,315 --> 00:33:37,675 on. 767 00:33:38,050 --> 00:33:40,570 I welcome you to, come on and tell me why I'm wrong. 768 00:33:40,600 --> 00:33:41,230 I just, 769 00:33:41,295 --> 00:33:41,415 I, 770 00:33:41,960 --> 00:33:43,250 know who will come back on, you 771 00:33:43,250 --> 00:33:43,700 know who, 772 00:33:43,775 --> 00:33:44,375 will come back on, 773 00:33:44,680 --> 00:33:45,340 what's his name? 774 00:33:45,450 --> 00:33:46,260 Bass Data guy. 775 00:33:47,565 --> 00:33:47,785 oh. 776 00:33:47,965 --> 00:33:50,685 Oh, are they flash 777 00:33:51,550 --> 00:33:51,840 Yeah, 778 00:33:52,515 --> 00:33:52,995 mark? 779 00:33:53,955 --> 00:33:54,795 No, sorry, Howard. 780 00:33:56,455 --> 00:33:56,860 Yeah. 781 00:33:57,290 --> 00:33:57,780 Fastest. 782 00:33:58,060 --> 00:33:58,720 Pure flash. 783 00:33:59,170 --> 00:33:59,410 Yeah. 784 00:33:59,875 --> 00:34:00,355 Yeah. 785 00:34:00,455 --> 00:34:01,035 all All right. 786 00:34:01,145 --> 00:34:03,795 yeah, Howard, you wanna tell me why I'm wrong? 787 00:34:03,895 --> 00:34:05,245 I'm more than happy to have you back. 788 00:34:05,605 --> 00:34:06,475 We can duke it out. 789 00:34:06,565 --> 00:34:07,075 We can duke it. 790 00:34:07,435 --> 00:34:08,486 wouldn't be the first time. 791 00:34:08,491 --> 00:34:10,525 Howard and I have disagreed on something. 792 00:34:10,765 --> 00:34:11,095 I don't know. 793 00:34:11,145 --> 00:34:14,955 it's just there are so many area, there are so many other places where I would 794 00:34:14,955 --> 00:34:16,625 wanna spend money in the backup system. 795 00:34:18,000 --> 00:34:18,120 Yep. 796 00:34:19,190 --> 00:34:19,191 but, 797 00:34:19,325 --> 00:34:20,915 comes down to what the cost is. 798 00:34:20,915 --> 00:34:23,555 If you could get flash down to a low enough point, 799 00:34:23,960 --> 00:34:24,500 which 800 00:34:24,545 --> 00:34:24,905 sense. 801 00:34:25,310 --> 00:34:26,090 of vast data, 802 00:34:26,615 --> 00:34:26,975 Yeah. 803 00:34:27,260 --> 00:34:32,070 architecture allows using flash in a, a significant way, 804 00:34:32,165 --> 00:34:32,735 that's why I brought 805 00:34:32,790 --> 00:34:33,315 close to cost. 806 00:34:33,565 --> 00:34:33,895 Okay. 807 00:34:34,105 --> 00:34:34,165 All 808 00:34:34,210 --> 00:34:34,420 Yeah. 809 00:34:34,465 --> 00:34:34,555 Okay. 810 00:34:35,185 --> 00:34:35,426 All right. 811 00:34:36,415 --> 00:34:37,135 All right. 812 00:34:37,945 --> 00:34:38,425 right. 813 00:34:38,995 --> 00:34:41,185 and then I got this whole other thing. 814 00:34:41,185 --> 00:34:42,415 I'm not gonna go into that other thing. 815 00:34:42,985 --> 00:34:52,516 but yeah, so d makes disk and cloud-based both physiologically feasible as 816 00:34:52,725 --> 00:34:54,825 well as economically feasible. 817 00:34:55,275 --> 00:34:55,605 is. 818 00:34:57,015 --> 00:35:04,335 Is there something that a person shopping for a dedupe system should be asking? 819 00:35:05,305 --> 00:35:05,515 Well, 820 00:35:05,575 --> 00:35:08,875 the important things that they should be asking in order to determine 821 00:35:09,085 --> 00:35:09,415 yeah, 822 00:35:09,415 --> 00:35:09,695 does it. 823 00:35:09,755 --> 00:35:10,685 great question. 824 00:35:10,715 --> 00:35:18,265 I think the question would be things about what's the restored performance? 825 00:35:18,265 --> 00:35:20,245 Because in the end, that's the only thing that matters. 826 00:35:20,695 --> 00:35:21,595 I remember. 827 00:35:22,675 --> 00:35:23,485 A product. 828 00:35:23,935 --> 00:35:27,155 Now, this product is still on the market, but I believe they 829 00:35:27,155 --> 00:35:28,865 have addressed this issue. 830 00:35:29,620 --> 00:35:31,415 remember a dedupe product. 831 00:35:31,475 --> 00:35:36,715 It was a Target dedupe product that had, I remember that had 400 megabytes 832 00:35:36,775 --> 00:35:40,405 a second throughput in an appliance. 833 00:35:40,885 --> 00:35:41,935 And like 10 megabits out 834 00:35:42,425 --> 00:35:44,835 it was 40, megabytes out. 835 00:35:45,165 --> 00:35:48,055 It had a 90%, what we call dedupe tax. 836 00:35:48,075 --> 00:35:48,215 Yep. 837 00:35:48,815 --> 00:35:51,665 the, because the problem with dedupe, depending on how you 838 00:35:51,665 --> 00:35:54,635 store it, is that you've got everything you need all over the 839 00:35:54,905 --> 00:35:55,865 All over the place. 840 00:35:55,870 --> 00:35:55,955 Yep. 841 00:35:56,645 --> 00:35:58,935 And so this was just a really bad design. 842 00:35:59,595 --> 00:36:04,275 And I believe that they addressed it and, because that product 843 00:36:04,275 --> 00:36:05,385 is still on the market today. 844 00:36:05,385 --> 00:36:08,805 But that version, one of that product was ble. 845 00:36:09,635 --> 00:36:12,005 so yeah, it's about restored performance, right? 846 00:36:12,005 --> 00:36:14,335 So one thing, oh, I'm. 847 00:36:14,635 --> 00:36:16,715 dedupe ratio is crap. 848 00:36:17,045 --> 00:36:18,515 Don't look at dedupe ratio. 849 00:36:18,515 --> 00:36:20,650 dedupe ratio is a made up number. 850 00:36:21,030 --> 00:36:23,550 I will, I'll go back to, I'll pick on Avamar. 851 00:36:23,555 --> 00:36:24,100 Avamar. 852 00:36:24,105 --> 00:36:25,810 Back in the day, they used to 853 00:36:25,810 --> 00:36:26,030 Oh, 854 00:36:26,140 --> 00:36:28,210 had a 400 to one DEDUP ratio. 855 00:36:28,300 --> 00:36:29,080 Do you remember this? 856 00:36:29,085 --> 00:36:29,470 Because 857 00:36:29,780 --> 00:36:30,460 I remember. 858 00:36:31,095 --> 00:36:32,770 every backup as a full backup. 859 00:36:32,770 --> 00:36:35,620 They're like, the way we store backups, which is the same way Druva stores 860 00:36:35,620 --> 00:36:37,630 backups, the way we store backups. 861 00:36:37,720 --> 00:36:40,300 It's even though they're incremental, it's like they're a full. 862 00:36:40,855 --> 00:36:41,125 , right? 863 00:36:41,130 --> 00:36:43,615 Because they behave a full during a restore. 864 00:36:44,035 --> 00:36:46,345 And so they considered every backup a full. 865 00:36:46,495 --> 00:36:49,645 And so they said, then therefore the dedup ratio is 400 to one. 866 00:36:49,645 --> 00:36:51,260 that was always complete nonsense. 867 00:36:51,750 --> 00:36:55,370 the other would be, I remember, again, I'm gonna pick on people equally. 868 00:36:55,375 --> 00:36:58,070 I remember sales reps of a certain large target. 869 00:36:59,315 --> 00:37:03,485 company that where you might've worked, where they would tell customers to go 870 00:37:03,485 --> 00:37:07,085 and do full backups more frequently because it made their dedup ratio better. 871 00:37:08,075 --> 00:37:10,505 , which is just, again, nonsense. 872 00:37:10,865 --> 00:37:18,855 What matters, in my opinion, what matters is how big is full backup versus 873 00:37:19,325 --> 00:37:22,145 how big are all the backups, right? 874 00:37:23,075 --> 00:37:24,065 if I have. 875 00:37:24,710 --> 00:37:26,900 If I, let me explain what I'm saying. 876 00:37:27,230 --> 00:37:31,470 If I have a hundred terabytes, if one full backup of my environment is a hundred 877 00:37:31,470 --> 00:37:37,470 terabytes and then after three months how big is, or whatever number you want. 878 00:37:38,140 --> 00:37:40,220 but it's just three months seems like a nice, long, 879 00:37:40,880 --> 00:37:41,035 what 880 00:37:41,795 --> 00:37:43,115 POC thing, 881 00:37:43,265 --> 00:37:43,535 Yep. 882 00:37:44,525 --> 00:37:47,625 after a hundred, after, three months, how. 883 00:37:48,795 --> 00:37:51,105 How much stuff is stored over there? 884 00:37:51,795 --> 00:37:52,635 That's what I'm saying. 885 00:37:52,635 --> 00:37:56,715 Don't dedupe ratios is nonsense that didn't come out in English. 886 00:37:56,875 --> 00:38:03,475 dedupe ratios are nonsense, but if I can fit a hundred terabytes right, if I have 887 00:38:03,480 --> 00:38:07,395 a hundred terabyte environment and then a series of incremental backups, and then 888 00:38:07,735 --> 00:38:10,165 over there, my question is how big is. 889 00:38:11,695 --> 00:38:13,195 How much data did 890 00:38:13,210 --> 00:38:13,870 Space. 891 00:38:14,710 --> 00:38:14,930 Yep. 892 00:38:14,965 --> 00:38:19,515 let's say it's 200 terabytes after 90 days. 893 00:38:19,725 --> 00:38:23,595 And then compare that with another product who writes a hundred terabytes? 894 00:38:23,925 --> 00:38:28,696 You backed up the same data, but you used half as much storage on the back end. 895 00:38:29,410 --> 00:38:29,630 . Yep. 896 00:38:29,640 --> 00:38:30,060 I'm saying. 897 00:38:30,280 --> 00:38:33,565 the problem is, and the other reason, and again, I'm a little extra 898 00:38:33,565 --> 00:38:35,485 sensitive to this cuz I work for Druva. 899 00:38:35,635 --> 00:38:37,365 People ask us what's our dedupe ratio? 900 00:38:37,385 --> 00:38:40,450 We're like, the thing is we're like the opposite of Avamar. 901 00:38:40,480 --> 00:38:44,080 we're actually similar to Avamar in that we're source I dedupe, 902 00:38:44,560 --> 00:38:46,220 but we don't use that funny math. 903 00:38:46,320 --> 00:38:47,100 So we could say 904 00:38:47,370 --> 00:38:47,650 Yeah. 905 00:38:48,520 --> 00:38:49,510 but that's nonsense. 906 00:38:49,840 --> 00:38:51,540 So you know, we say, we. 907 00:38:52,315 --> 00:38:54,865 because we also do incremental forever backups. 908 00:38:55,475 --> 00:38:56,015 the problem. 909 00:38:56,065 --> 00:39:02,155 but I know that on average, if we have a hundred terabyte customer, we store, 910 00:39:02,545 --> 00:39:06,415 roughly a year's worth of backups in less than a hundred terabytes of disk. 911 00:39:06,515 --> 00:39:06,730 Yeah. 912 00:39:06,775 --> 00:39:07,125 Right. 913 00:39:07,150 --> 00:39:12,070 And I think it's important there to also account for that increment, like 914 00:39:12,250 --> 00:39:14,650 how I look at these like numbers. 915 00:39:14,710 --> 00:39:17,380 I totally get what you said, Curtis, like you should just do an apples. 916 00:39:17,380 --> 00:39:17,650 But 917 00:39:17,725 --> 00:39:18,085 Yeah. 918 00:39:18,130 --> 00:39:20,530 don't have that ability, you should also look to say, okay, 919 00:39:21,010 --> 00:39:22,720 I have a hundred terabyte full. 920 00:39:22,930 --> 00:39:24,761 And then say, my daily change rate is 2%. 921 00:39:25,880 --> 00:39:26,095 right? 922 00:39:26,095 --> 00:39:28,225 So if I do 2% for a month, right? 923 00:39:28,225 --> 00:39:30,835 That's, what is that two 60? 924 00:39:31,855 --> 00:39:33,235 60 more terabytes, right? 925 00:39:33,235 --> 00:39:37,585 So it should be 160 terabytes worth of data that I sent over, right? 926 00:39:37,585 --> 00:39:41,035 For 160 terabytes worth of data, how much should I actually store? 927 00:39:42,335 --> 00:39:42,740 Yeah. 928 00:39:42,815 --> 00:39:44,545 you similar things to what you're saying, right? 929 00:39:45,045 --> 00:39:45,265 But 930 00:39:45,380 --> 00:39:45,620 The, 931 00:39:45,845 --> 00:39:48,665 because what you're saying is if you had the two products, then 932 00:39:48,670 --> 00:39:49,835 you could do a direct comparison. 933 00:39:49,835 --> 00:39:52,205 But I'm saying if you don't have the two products, then 934 00:39:52,205 --> 00:39:53,435 here's another way you could 935 00:39:53,465 --> 00:39:57,195 I, I would argue that there's no way to compare them if you 936 00:39:57,195 --> 00:39:58,240 don't have two pro, if you're 937 00:39:58,295 --> 00:39:58,585 Yeah. 938 00:39:59,080 --> 00:40:01,450 if you're not doing a true comparison. 939 00:40:01,550 --> 00:40:01,670 A 940 00:40:01,775 --> 00:40:03,575 it's just that d math is funny, right? 941 00:40:04,790 --> 00:40:07,190 So different products charge differently, right? 942 00:40:07,190 --> 00:40:07,640 You look 943 00:40:07,770 --> 00:40:08,060 Yeah. 944 00:40:08,250 --> 00:40:12,780 like when you look at Metallic, which competes with Druva, they have a frontend 945 00:40:12,840 --> 00:40:14,520 price and we have a backend price. 946 00:40:14,800 --> 00:40:15,090 Yeah. 947 00:40:15,270 --> 00:40:17,700 They actually have the front end price, and then you also need 948 00:40:17,700 --> 00:40:19,050 to pay for the backend storage. 949 00:40:19,230 --> 00:40:21,300 So you're paying, so how do you compare that? 950 00:40:21,970 --> 00:40:23,950 it's just, it's difficult 951 00:40:23,995 --> 00:40:24,505 hard. 952 00:40:24,565 --> 00:40:24,925 Yeah. 953 00:40:25,060 --> 00:40:25,840 it's hard. 954 00:40:26,150 --> 00:40:29,720 but all I'm saying is dedup ratio is crap and 955 00:40:29,795 --> 00:40:30,185 Yep, 956 00:40:30,650 --> 00:40:31,280 anything. 957 00:40:31,730 --> 00:40:35,360 but what does matter is how much data are you storing on that backend because 958 00:40:35,360 --> 00:40:37,400 you will be paying for that one way 959 00:40:37,445 --> 00:40:37,665 Yep. 960 00:40:39,035 --> 00:40:39,515 All right. 961 00:40:40,265 --> 00:40:44,735 I don't know if we made this, if we, if this is clear as mud or what, but, 962 00:40:44,785 --> 00:40:48,995 I hope that was helpful and, maybe we ticked off Howard and Howard's 963 00:40:48,995 --> 00:40:51,155 gonna come on next week's episode. 964 00:40:51,335 --> 00:40:51,695 . I dunno. 965 00:40:54,155 --> 00:40:54,275 All 966 00:40:54,410 --> 00:40:54,800 join us, 967 00:40:54,800 --> 00:40:55,430 Howard, 968 00:40:55,455 --> 00:40:58,345 thanks for, thanks for helping me with my network as well, 969 00:40:58,705 --> 00:40:59,395 anytime, Curtis. 970 00:41:00,205 --> 00:41:02,155 Just remember I am not tech support. 971 00:41:03,640 --> 00:41:04,045 Yeah. 972 00:41:04,225 --> 00:41:04,645 Yeah. 973 00:41:05,185 --> 00:41:05,725 All right. 974 00:41:05,885 --> 00:41:08,675 and thanks to the listeners and remember to subscribe. 975 00:41:11,587 --> 00:41:16,287 The Backup Wrap Up is written, recorded, and produced by me, W. Curtis Preston. 976 00:41:16,877 --> 00:41:21,637 If you need backup or DR consulting, content generation, or expert witness 977 00:41:21,637 --> 00:41:24,437 work, check out backupcentral.com. 978 00:41:24,947 --> 00:41:28,007 You can also find links for my O'Reilly books on the same website. 979 00:41:28,737 --> 00:41:32,707 Remember, this is an independent podcast, and any opinions that 980 00:41:32,707 --> 00:41:36,677 you hear are those of the speaker and not necessarily an employer. 981 00:41:37,547 --> 00:41:38,207 Thanks for listening