1 00:00:00,227 --> 00:00:04,457 You found the backup wrap up your go-to podcast for all things 2 00:00:04,457 --> 00:00:07,157 backup recovery and cyber recovery. 3 00:00:07,787 --> 00:00:11,267 In this episode, we challenge the notion that cloud versus 4 00:00:11,267 --> 00:00:13,787 tape is an either or proposition. 5 00:00:14,387 --> 00:00:16,727 I saw this LinkedIn post claiming that. 6 00:00:17,072 --> 00:00:21,422 LTO Library Robots are the only robots not making things easier 7 00:00:21,422 --> 00:00:24,092 in 2025, and it got my attention. 8 00:00:24,572 --> 00:00:29,672 He suggested that tape is too slow and unreliable compared to cloud for active 9 00:00:29,672 --> 00:00:32,162 archives, which got me a little worked up. 10 00:00:32,432 --> 00:00:37,142 Here's the thing, if you're putting data in cloud's, cheaper tiers, guess what? 11 00:00:37,472 --> 00:00:39,452 It's probably sitting on tape anyway. 12 00:00:39,947 --> 00:00:45,047 Persona and I break down, uh, I think a real comparison, cost, access, time, data 13 00:00:45,047 --> 00:00:47,507 integrity, and location considerations. 14 00:00:47,867 --> 00:00:52,607 But honestly, I just start by rejecting the premise of the question because it's 15 00:00:52,607 --> 00:00:54,527 probably on tape if it's in the cloud. 16 00:00:55,007 --> 00:00:56,297 I hope you enjoy the episode. 17 00:00:56,777 --> 00:01:00,917 By the way, if you don't know who I am, I'm w Curtis Preston, AKA, Mr. 18 00:01:00,917 --> 00:01:05,237 Backup, and I've been passionate about backup and recovery for over 30 years, 19 00:01:05,507 --> 00:01:07,157 ever since I had to tell my boss. 20 00:01:07,467 --> 00:01:10,527 We had no backups of that database that we just lost. 21 00:01:11,067 --> 00:01:14,127 I don't want that to happen to you, and that's why I do this podcast. 22 00:01:14,667 --> 00:01:19,917 On this podcast, we turn unappreciated backup admins into Cyber Recovery Heroes. 23 00:01:20,157 --> 00:01:22,287 This is the backup wrap up. 24 00:01:36,265 --> 00:01:37,285 Welcome to the show. 25 00:01:38,395 --> 00:01:43,105 Hi, I am w Curtis Preston, AKA, Mr. Backup, and I have with me a guy who has 26 00:01:43,105 --> 00:01:45,985 been flaunting his new job in my face. 27 00:01:46,985 --> 00:01:49,225 Prasanna Malaiyandi, how's it going? 28 00:01:49,225 --> 00:01:49,765 Prasanna I. 29 00:01:50,085 --> 00:01:51,370 I am good Curtis. 30 00:01:51,370 --> 00:01:55,630 I, I wouldn't call it flaunting, but I'm just excited about my 31 00:01:55,630 --> 00:01:57,190 new job. 32 00:01:57,505 --> 00:01:59,065 Oh, you should see the campus. 33 00:01:59,245 --> 00:02:00,145 It's so amazing. 34 00:02:00,355 --> 00:02:01,195 And the coffee. 35 00:02:01,405 --> 00:02:04,795 And the coffee and the food and the food choices are so great. 36 00:02:04,855 --> 00:02:07,735 And I'm like, I, my coffee 37 00:02:08,080 --> 00:02:08,350 You. 38 00:02:08,350 --> 00:02:10,750 Yeah, I see the, I see your coffee choice in the corner. 39 00:02:12,325 --> 00:02:12,685 Yeah. 40 00:02:12,865 --> 00:02:14,245 Let's see if I tilt the camera. 41 00:02:14,485 --> 00:02:17,575 That's my new espresso machine, and you can see at the bottom. 42 00:02:18,790 --> 00:02:22,240 For those of you watching on YouTube, it has a like a little shelf 43 00:02:22,240 --> 00:02:24,490 with a drawer of coffee choices. 44 00:02:24,505 --> 00:02:30,295 And as you if you, and so I think people you should watch, or listeners, you 45 00:02:30,295 --> 00:02:32,665 should watch the podcast on YouTube 46 00:02:32,995 --> 00:02:38,575 and keep an eye on that shelf and see how many go missing every single time 47 00:02:38,575 --> 00:02:44,845 you watch between episodes to see how big of a caffeine habit cur Curtis has. 48 00:02:45,625 --> 00:02:45,915 yeah. 49 00:02:45,920 --> 00:02:46,420 Yeah. 50 00:02:47,470 --> 00:02:53,320 Curtis definitely has a caffeine habit, but the, the rules are that if you 51 00:02:53,320 --> 00:02:57,730 get coffee like at like, you know, a breakfast place and then you come home, 52 00:02:58,000 --> 00:02:59,950 those coffees, they don't, they don't count. 53 00:02:59,950 --> 00:03:00,670 They're together. 54 00:03:00,670 --> 00:03:01,960 You can't count them together. 55 00:03:02,650 --> 00:03:04,030 'cause if you did, that would be a lot. 56 00:03:06,345 --> 00:03:09,370 Have you ever thought about injecting coffee directly into your veins? 57 00:03:11,205 --> 00:03:12,730 I don't think that would work out well. 58 00:03:14,770 --> 00:03:18,040 So today's topic is not coffee. 59 00:03:18,130 --> 00:03:24,445 Today's topic is about tape versus cloud and the, the or, or 60 00:03:24,565 --> 00:03:26,485 there, why, why, why is it versus. 61 00:03:27,460 --> 00:03:27,670 Because 62 00:03:27,670 --> 00:03:31,420 in the last epi, last episode, we talked about how the 63 00:03:31,420 --> 00:03:33,220 cloud is probably using tape. 64 00:03:35,080 --> 00:03:37,180 Don't, don't, don't be jumping to the punchline, 65 00:03:37,330 --> 00:03:41,435 No, I'm, no, I'm just restating You said cloud versus tape 66 00:03:42,100 --> 00:03:42,520 Yes. 67 00:03:42,520 --> 00:03:45,670 That is the name of the, that is the name of the show, of the episode. 68 00:03:47,110 --> 00:03:50,200 And the reason, the reason it's cloud versus tape is that all of 69 00:03:50,200 --> 00:03:51,970 this started with one guy's post. 70 00:03:52,015 --> 00:03:52,305 okay. 71 00:03:52,690 --> 00:03:53,230 Right? 72 00:03:53,905 --> 00:03:58,255 So this all started from this LinkedIn post that I saw and. 73 00:03:58,960 --> 00:04:04,000 It, it had a headline that, you know, a good headline in 74 00:04:04,000 --> 00:04:05,530 a post is, is solid, right? 75 00:04:05,920 --> 00:04:09,550 And this PO and the headline of this post caught me, right? 76 00:04:09,970 --> 00:04:10,990 And here's what it said. 77 00:04:11,110 --> 00:04:13,210 This is, this is, this answers your question. 78 00:04:13,280 --> 00:04:15,620 Why is this a versus episode? 79 00:04:17,210 --> 00:04:22,490 Here's what it said, LTO Library robots, the only robots not 80 00:04:22,490 --> 00:04:24,260 making things easier in 2025. 81 00:04:25,310 --> 00:04:25,600 What. 82 00:04:27,170 --> 00:04:28,070 That's what it says. 83 00:04:28,250 --> 00:04:29,660 And he says Controversial. 84 00:04:29,780 --> 00:04:30,380 Nope. 85 00:04:30,860 --> 00:04:36,500 Uh, and then he goes on to basically do a tape versus cloud, uh, post. 86 00:04:36,875 --> 00:04:40,955 and, and we'll include, And we'll include the link to the post in the show notes. 87 00:04:41,450 --> 00:04:41,870 Yeah. 88 00:04:41,870 --> 00:04:45,050 So all you tape, well, actually I don't, I don't know. 89 00:04:45,410 --> 00:04:47,540 Just 'cause you're backup friendly doesn't mean you're tape friendly. 90 00:04:47,900 --> 00:04:50,390 You should be tape friendly if you're listening to, if you listen to this 91 00:04:50,480 --> 00:04:50,700 No. 92 00:04:50,700 --> 00:04:53,315 Just so they can see what the original post was about. 93 00:04:53,660 --> 00:04:54,380 Yeah, yeah, yeah. 94 00:04:54,410 --> 00:04:56,090 No, I want him, I want him to go comment on, 95 00:04:58,910 --> 00:05:02,510 um, you know, I mean, he'll, he'll like it. 96 00:05:02,510 --> 00:05:05,720 'cause basically the pose is, you know, advertising how amazing they 97 00:05:05,720 --> 00:05:14,860 are first off, he makes two, um, I'll say claims about tape, one of which 98 00:05:15,190 --> 00:05:17,410 I thought was totally outta line. 99 00:05:17,500 --> 00:05:21,490 And then the other I think is in the context of, you know, in 100 00:05:21,490 --> 00:05:26,590 the context is not that big of a deal, but when we talk about. 101 00:05:28,565 --> 00:05:33,575 So his points, so he's basically suggesting, first off, he's saying 102 00:05:33,575 --> 00:05:38,825 cloud is better than tape for the concept of an active archive. 103 00:05:38,855 --> 00:05:40,325 Let me define active archive. 104 00:05:40,325 --> 00:05:40,985 And that is. 105 00:05:41,825 --> 00:05:45,905 Well, it's somewhat self-defining, but it, you know, a traditional archive 106 00:05:45,905 --> 00:05:49,115 is you make a copy and you put it on a shelf and you hope you never touch it. 107 00:05:49,115 --> 00:05:53,105 An active archive is secondary data, right. 108 00:05:53,105 --> 00:05:56,585 You know, uh, maybe even tertiary, you know, data of 109 00:05:56,585 --> 00:05:58,445 secondary or tertiary importance. 110 00:05:59,045 --> 00:06:02,975 But we might occasionally access it. 111 00:06:03,005 --> 00:06:04,535 Not rarely, but occasionally. 112 00:06:05,540 --> 00:06:05,840 And 113 00:06:06,500 --> 00:06:11,000 just to chime in on that, so when you said active archive, I was like, that 114 00:06:11,000 --> 00:06:14,240 sounds like an oxymoron, you know? 115 00:06:15,050 --> 00:06:15,320 Right. 116 00:06:15,320 --> 00:06:19,610 Because no one thinks active and archive together now having 117 00:06:19,610 --> 00:06:20,990 worked in storage for a long time. 118 00:06:20,990 --> 00:06:21,230 Right. 119 00:06:21,230 --> 00:06:24,650 Normally what we'd also call that is HSM, right. 120 00:06:24,650 --> 00:06:26,810 Hierarchical storage management. 121 00:06:26,855 --> 00:06:27,485 yes. 122 00:06:27,485 --> 00:06:27,875 Yes. 123 00:06:27,875 --> 00:06:29,740 I, I think HSM would qualify Yeah. 124 00:06:29,945 --> 00:06:36,095 Where older data sort of gets tiered off into lower costs, lower storage, 125 00:06:36,365 --> 00:06:38,765 just to keep it still visible, right. 126 00:06:38,765 --> 00:06:39,665 And accessible. 127 00:06:39,665 --> 00:06:44,105 But there is sort of a penalty of being able to read that data back. 128 00:06:44,105 --> 00:06:46,955 But from a client perspective, it's still available and usable. 129 00:06:48,110 --> 00:06:50,840 Right, and it, it. 130 00:06:51,140 --> 00:06:53,960 It might not be fully HSM, right? 131 00:06:53,990 --> 00:06:57,830 'cause the, the, the idea of HSM is that it, it just magically comes back. 132 00:06:57,830 --> 00:07:02,060 But, but it, it is, it is this idea of an archive, which is actively, 133 00:07:02,120 --> 00:07:06,110 that you're actively pulling stuff out on a regular basis. 134 00:07:06,530 --> 00:07:10,430 And as opposed to sort of the traditional, like you said, like 135 00:07:10,430 --> 00:07:13,640 true archive where you put it there and you just, you never access it. 136 00:07:13,710 --> 00:07:16,330 and just, uh, one example, and 137 00:07:16,715 --> 00:07:19,595 let me know if you think this makes sense for our listeners too, is like. 138 00:07:20,345 --> 00:07:24,605 Medical research like DNA research, like these models and this data, 139 00:07:24,605 --> 00:07:26,885 scientific data that they collect, right? 140 00:07:26,885 --> 00:07:28,925 They may not actively need it, right? 141 00:07:28,925 --> 00:07:30,575 To use it on a day-to-day basis. 142 00:07:30,575 --> 00:07:33,935 They might need to store it because in some future point they may need 143 00:07:33,935 --> 00:07:35,255 to be able to reference that data. 144 00:07:35,255 --> 00:07:38,885 And usually this data is very, very large data sets. 145 00:07:39,305 --> 00:07:42,875 And so you don't wanna necessarily keep it stored on your primary 146 00:07:42,875 --> 00:07:45,185 storage system because of cost. 147 00:07:46,975 --> 00:07:47,195 Yes. 148 00:07:48,110 --> 00:07:49,490 Um, but you still want it. 149 00:07:49,490 --> 00:07:51,920 You wanna make sure it stays around and you wanna be able to 150 00:07:51,920 --> 00:07:53,900 access it right there, there. 151 00:07:53,900 --> 00:07:58,310 By the way, there's an entire, uh, group called the Active Archive Alliance, which 152 00:07:58,310 --> 00:08:03,260 is a whole, whole number of people, some of which listen to the show that, um, that 153 00:08:03,260 --> 00:08:07,760 are basically vendors dedicated to this concept of the, of the active archive. 154 00:08:09,140 --> 00:08:10,730 So he, you know. 155 00:08:11,765 --> 00:08:15,815 He was basically suggesting that if you have an active archive, if you're going 156 00:08:15,815 --> 00:08:21,485 to regularly retrieve data from the archive, then again, according to the 157 00:08:21,485 --> 00:08:26,255 post, you should put it in the cloud and not on tape because cloud is better and 158 00:08:26,255 --> 00:08:29,345 tape is, his two claims are too slow. 159 00:08:29,945 --> 00:08:34,055 And he basically implied that it was unreliable. 160 00:08:34,055 --> 00:08:34,385 Right? 161 00:08:34,385 --> 00:08:38,135 He said, if you want that clip at your fingertips without the weight 162 00:08:38,135 --> 00:08:40,505 and the will it load anxiety. 163 00:08:41,225 --> 00:08:43,175 I was like, what the, what, what is that? 164 00:08:43,445 --> 00:08:46,445 So, so basically he's saying it takes too long and, and it's, and 165 00:08:46,445 --> 00:08:50,675 it's not reliable and so therefore you need to put it in the cloud and. 166 00:08:51,890 --> 00:08:52,220 okay. 167 00:08:52,220 --> 00:08:52,730 Keep going. 168 00:08:53,285 --> 00:08:53,615 Yeah. 169 00:08:53,615 --> 00:08:55,805 Well, but that's, that was the point of his post. 170 00:08:56,015 --> 00:08:56,525 Okay. 171 00:08:57,455 --> 00:09:02,315 And I, I, I know that, that one of the first things that's going to come 172 00:09:02,315 --> 00:09:05,765 to your mind is if you put it in the cloud, it might indeed be on tape. 173 00:09:06,305 --> 00:09:06,695 Right? 174 00:09:06,935 --> 00:09:10,385 So it's, so first off, there, there is no such thing as tape versus cloud. 175 00:09:10,895 --> 00:09:12,755 It's, it's really tape versus. 176 00:09:14,840 --> 00:09:16,640 His post was about T versus cloud. 177 00:09:16,640 --> 00:09:21,500 And so I thought it was a, you know, a good topic to discuss and also to just 178 00:09:21,500 --> 00:09:28,970 bring up the fact that, um, if, if the listeners don't know, if you are using 179 00:09:29,300 --> 00:09:35,780 the lower cost tiers of cloud storage, you are most likely storing your data on tape. 180 00:09:36,020 --> 00:09:38,960 Go listen to our episode a couple or a couple episodes ago. 181 00:09:38,960 --> 00:09:39,890 We just talked about this. 182 00:09:40,640 --> 00:09:43,250 Did we, I forgot. 183 00:09:43,495 --> 00:09:44,195 Yes, we did. 184 00:09:44,195 --> 00:09:44,555 Curtis, 185 00:09:46,040 --> 00:09:48,140 Do you remember what the title of that episode was? 186 00:09:48,225 --> 00:09:50,510 I don't think it's been published yet, but it, 187 00:09:50,510 --> 00:09:51,620 Oh, is that what it is? 188 00:09:51,920 --> 00:09:52,310 Okay. 189 00:09:53,015 --> 00:09:53,645 We did 190 00:09:53,660 --> 00:09:54,020 Oh, yeah. 191 00:09:54,020 --> 00:09:55,610 We did, we did talk about what, 192 00:09:55,895 --> 00:09:57,140 IS tape backup dead. 193 00:09:58,220 --> 00:09:59,180 Oh, yeah, yeah. 194 00:09:59,300 --> 00:10:00,530 So yes. 195 00:10:00,650 --> 00:10:00,950 Right. 196 00:10:00,980 --> 00:10:04,070 So we did, we did mention that that in the tape back is taped back up 197 00:10:04,070 --> 00:10:07,370 dead episode, which is, which is, which is related to this episode. 198 00:10:08,285 --> 00:10:11,015 And there, and then we made a lot of really good points in that 199 00:10:11,015 --> 00:10:15,185 episode, which we're not gonna take the time to rehash in this episode. 200 00:10:15,245 --> 00:10:19,715 This is mainly about, first off, this idea of is cloud better than tape? 201 00:10:19,715 --> 00:10:23,585 And first off, I'm gonna say that I reject the premise because it's 202 00:10:23,585 --> 00:10:26,705 really disc versus tape, right? 203 00:10:26,765 --> 00:10:28,205 Not cloud versus 204 00:10:28,520 --> 00:10:33,950 but even if I just took, so I think that is definitely a fair angle to look at it 205 00:10:34,250 --> 00:10:34,970 in my mind. 206 00:10:34,970 --> 00:10:38,810 When you said cloud versus tape, the one thing that, or couple 207 00:10:38,810 --> 00:10:39,980 things pop to mind, right? 208 00:10:40,040 --> 00:10:42,320 So one is. 209 00:10:43,205 --> 00:10:45,065 Storing data on cloud. 210 00:10:45,065 --> 00:10:49,475 If your infrastructure that you need to access the data is on premises, 211 00:10:50,705 --> 00:10:54,665 you're going to need to pull the data back and send the data up. 212 00:10:55,145 --> 00:10:58,775 And that's gonna take time to, especially if these are active archive 213 00:10:59,075 --> 00:11:02,315 scenarios, like I had mentioned, where you might have a very large data set. 214 00:11:03,065 --> 00:11:03,335 Right? 215 00:11:03,335 --> 00:11:03,515 So 216 00:11:03,515 --> 00:11:06,935 one is it's not gonna be instantaneous, right? 217 00:11:06,935 --> 00:11:09,995 Versus something that might be on premises like your tape library 218 00:11:10,925 --> 00:11:12,665 Yeah, that's a, that's a great point. 219 00:11:12,665 --> 00:11:12,875 Right? 220 00:11:12,875 --> 00:11:17,405 So, so if we're, if we are talking like, if we, if we don't reject the premise and 221 00:11:17,405 --> 00:11:22,985 we say cloud versus tape, like setting aside the fact that cloud is often tape, 222 00:11:23,855 --> 00:11:24,185 right? 223 00:11:24,875 --> 00:11:27,455 If we're talking about putting your data in a cloud vendor, ver 224 00:11:27,545 --> 00:11:31,595 put, versus putting your data in a tape library that you control Yes. 225 00:11:31,655 --> 00:11:32,045 Access. 226 00:11:33,995 --> 00:11:37,115 Could be an issue depending on the size of the things that we're talking about. 227 00:11:37,115 --> 00:11:40,385 It's not instantaneous to move things back and forth from the cloud. 228 00:11:41,345 --> 00:11:45,995 Um, potentially this is mitigated by moving where your infrastructure is. 229 00:11:45,995 --> 00:11:49,025 If your infrastructure is up, you know, if you're, if you're storing the data 230 00:11:49,025 --> 00:11:53,585 in a AWS and your compute happens to be an AWS, it's a much shorter trip. 231 00:11:53,795 --> 00:11:54,005 Yep. 232 00:11:54,275 --> 00:11:54,935 Um, 233 00:11:55,535 --> 00:11:55,805 the, 234 00:11:56,015 --> 00:11:56,135 I 235 00:11:56,345 --> 00:11:57,830 that, that is a really valid point. 236 00:11:58,115 --> 00:12:00,005 And then the other, so I had two more points. 237 00:12:00,695 --> 00:12:04,055 Um, the second one is cost, right? 238 00:12:04,055 --> 00:12:06,215 He doesn't necessarily talk about that, but. 239 00:12:06,605 --> 00:12:07,865 He doesn't talk about cost at 240 00:12:08,165 --> 00:12:12,665 Right, but if, say you wanted not the very lowest cost storage 241 00:12:12,665 --> 00:12:16,835 that's in the cloud, but something that's a little bit more practical, 242 00:12:16,925 --> 00:12:17,375 right? 243 00:12:17,795 --> 00:12:19,985 You're paying per month, per 244 00:12:19,985 --> 00:12:21,335 gigabyte, right? 245 00:12:21,515 --> 00:12:25,025 If you have very, very large data sets and multiple, very, very large 246 00:12:25,025 --> 00:12:29,705 data sets, the cost can quickly add up, especially if this is archive, 247 00:12:30,035 --> 00:12:33,275 which implies that you're keeping it for a very, very, very long time. 248 00:12:34,025 --> 00:12:34,505 Yes. 249 00:12:35,705 --> 00:12:37,115 And that's a really good point, right? 250 00:12:37,115 --> 00:12:42,965 And, and that's been historically my, you know, when I'll, I'll quickly 251 00:12:42,965 --> 00:12:47,315 summarize the things that I, um, talked about in the other episode. 252 00:12:47,675 --> 00:12:52,115 Cost, uh, initial integrity and long-term integrity, right? 253 00:12:52,115 --> 00:12:52,505 So. 254 00:12:53,060 --> 00:12:57,500 Tape is much, much, much, much, much cheaper than literally 255 00:12:57,560 --> 00:12:59,240 in all three categories. 256 00:12:59,240 --> 00:13:02,900 It is at least one, sometimes two, sometimes three orders of 257 00:13:02,900 --> 00:13:05,780 magnitude better in that category. 258 00:13:06,110 --> 00:13:10,280 So, and, and in fact, I've, I've regular, I've regularly made the 259 00:13:10,280 --> 00:13:16,190 claim if what we're talking about is long-term storage, that set aside, 260 00:13:16,190 --> 00:13:17,870 set, set aside, cloud for the minute. 261 00:13:18,425 --> 00:13:23,075 Even if we looked at disc that we're going to own, which we know long term is going 262 00:13:23,075 --> 00:13:27,785 to be cheaper than doing it in the cloud, uh, assuming we have a hundred percent 263 00:13:27,785 --> 00:13:28,985 utilization and all of that, right? 264 00:13:28,985 --> 00:13:34,715 But, um, if the disc were free, 265 00:13:36,785 --> 00:13:39,875 right, it would still be more expensive than tape. 266 00:13:39,910 --> 00:13:40,120 Yep. 267 00:13:40,895 --> 00:13:41,165 Why? 268 00:13:41,195 --> 00:13:42,245 Why would I say that? 269 00:13:42,370 --> 00:13:43,420 Power cooling 270 00:13:44,195 --> 00:13:45,185 Power and cooling 271 00:13:45,725 --> 00:13:46,145 right? 272 00:13:46,145 --> 00:13:47,345 And data center space. 273 00:13:47,440 --> 00:13:47,650 Yep. 274 00:13:48,980 --> 00:13:54,290 Um, because it's going to be bigger than the tape, it's going to create way 275 00:13:54,290 --> 00:14:00,170 more power need, or it's gonna have way more power needs, which then creates 276 00:14:00,170 --> 00:14:04,850 way more heat, which then has to be dissipated through, uh, cooling, right? 277 00:14:05,360 --> 00:14:09,830 So that's why I've often made the, the argument, again, cloud is a different 278 00:14:09,830 --> 00:14:14,840 argument, but when it comes to cost, even if disc were free tape would be cheaper. 279 00:14:15,530 --> 00:14:15,710 Why? 280 00:14:15,740 --> 00:14:16,490 Why is that? 281 00:14:17,255 --> 00:14:17,795 because. 282 00:14:18,935 --> 00:14:20,135 Well, you just talked about it, right? 283 00:14:20,135 --> 00:14:23,075 Because tape is, you write it and then it just sits there. 284 00:14:23,075 --> 00:14:23,285 Right? 285 00:14:23,285 --> 00:14:24,365 You are not powering it. 286 00:14:24,365 --> 00:14:25,295 You're not cooling it. 287 00:14:25,295 --> 00:14:25,625 Right? 288 00:14:25,925 --> 00:14:28,865 In fact, it should be, I think in the last episode we talked about, right? 289 00:14:29,045 --> 00:14:30,965 It should be stored at room temperature, Right. 290 00:14:31,565 --> 00:14:32,135 So 291 00:14:32,915 --> 00:14:39,185 And, and the, and the even and the drives, unlike disc, the drives aren't consuming 292 00:14:39,185 --> 00:14:42,125 much power when they're not being used. 293 00:14:42,395 --> 00:14:42,845 Right. 294 00:14:43,265 --> 00:14:44,525 Um, and they're consuming. 295 00:14:44,585 --> 00:14:47,795 And, and when the drive is doing something, it's consuming less 296 00:14:47,795 --> 00:14:49,865 power than the same amount of disc. 297 00:14:49,895 --> 00:14:50,255 Right. 298 00:14:50,855 --> 00:14:53,705 Um, the tapes consume zero power, right? 299 00:14:53,705 --> 00:14:56,255 So the, the tape li a tape library that is. 300 00:14:57,125 --> 00:15:01,535 You know, this big and a tape library that's this big, consumes 301 00:15:01,535 --> 00:15:02,435 the same amount of power. 302 00:15:02,570 --> 00:15:08,090 And, and for people, uh, listening to the podcast, Curtis basically showed 303 00:15:08,090 --> 00:15:10,640 maybe a foot and a half versus six feet. 304 00:15:11,375 --> 00:15:11,615 Yeah. 305 00:15:11,615 --> 00:15:12,215 There you go. 306 00:15:12,605 --> 00:15:13,295 Um, 307 00:15:13,685 --> 00:15:13,745 So, 308 00:15:14,435 --> 00:15:14,795 so, 309 00:15:14,885 --> 00:15:15,185 yes. 310 00:15:15,245 --> 00:15:17,075 those listening at home, yeah. 311 00:15:17,255 --> 00:15:18,485 Uh, yeah, go ahead. 312 00:15:19,265 --> 00:15:19,595 Nope. 313 00:15:20,600 --> 00:15:26,030 Well, it just, it just cost is definitely the number one, uh, thing. 314 00:15:26,030 --> 00:15:32,510 Because, because then when we talk about cloud, your point is super, super valid 315 00:15:32,510 --> 00:15:35,990 because if we're talking cloud, you are paying every single month forever. 316 00:15:35,990 --> 00:15:39,050 As long as you're keeping that, that archive in there and you're 317 00:15:39,050 --> 00:15:46,250 paying by the gigabyte and that per gigabyte price is, um, that 318 00:15:46,250 --> 00:15:48,170 you're paying every month is likely. 319 00:15:49,460 --> 00:15:53,060 And it's probably similar to the acquisition price of tape. 320 00:15:54,020 --> 00:15:54,350 Right? 321 00:15:54,680 --> 00:15:54,980 Right. 322 00:15:54,980 --> 00:15:57,260 Meaning, meaning just buying that one tape and you're, 323 00:15:57,320 --> 00:15:58,400 you're gonna pay that per month. 324 00:15:58,580 --> 00:16:02,180 and then the other thing to also mention is active. 325 00:16:02,210 --> 00:16:04,220 So you talked about active archive, so I'm 326 00:16:04,220 --> 00:16:07,070 assuming you are gonna be reading the data periodically, 327 00:16:07,475 --> 00:16:07,825 Right. 328 00:16:07,970 --> 00:16:08,270 right? 329 00:16:08,270 --> 00:16:13,700 Which means that depending on what storage tier or storage class you use 330 00:16:13,700 --> 00:16:18,890 for the cloud, you may have different costs for reading the data back. 331 00:16:19,715 --> 00:16:20,705 Right, which 332 00:16:20,705 --> 00:16:24,935 could be very very expensive for the lowest cost tier of storage. 333 00:16:25,535 --> 00:16:26,645 right, right. 334 00:16:26,645 --> 00:16:26,795 Yeah. 335 00:16:26,795 --> 00:16:28,205 They do charge you quite a bit. 336 00:16:28,235 --> 00:16:33,365 'cause the idea with of depending on the class that you put in, so your choice 337 00:16:33,365 --> 00:16:35,412 is paying more per gigabyte per month. 338 00:16:36,417 --> 00:16:38,337 And be able to access it no problem. 339 00:16:38,337 --> 00:16:42,147 Or pay less per gigabyte per month and then pay a lot when you access it. 340 00:16:42,957 --> 00:16:44,757 Um, you know, pick your poison, right. 341 00:16:45,297 --> 00:16:53,517 Um, and also the cost per month is also going to depend, usually depends on how 342 00:16:53,517 --> 00:16:55,647 quickly you'll be able to access it. 343 00:16:56,007 --> 00:17:00,657 At least it isn't a WSI know like in Google storage, Google Cloud storage. 344 00:17:01,002 --> 00:17:02,022 the same, I think 345 00:17:02,187 --> 00:17:02,517 it's the 346 00:17:02,517 --> 00:17:06,387 same, the same access time, but they just charge you again 347 00:17:06,417 --> 00:17:08,097 based on when you access it. 348 00:17:08,112 --> 00:17:08,382 Yep. 349 00:17:08,982 --> 00:17:10,362 Uh, I, I have one more. 350 00:17:11,217 --> 00:17:11,637 Yeah. 351 00:17:11,727 --> 00:17:13,677 Okay, so the third one, right? 352 00:17:13,947 --> 00:17:16,677 My third point about this, just cloud versus tape, 353 00:17:17,517 --> 00:17:19,977 assuming once again that you're running on premises, 354 00:17:20,757 --> 00:17:21,147 right? 355 00:17:21,507 --> 00:17:24,027 And you're storing your data in the cloud, right? 356 00:17:24,027 --> 00:17:27,387 We talked about the cost of storing it up there, the cost of the time 357 00:17:27,387 --> 00:17:29,217 it takes to transfer it up and down. 358 00:17:29,322 --> 00:17:29,742 Mm-hmm. 359 00:17:30,057 --> 00:17:33,387 One other cost we didn't talk about is egress cost, 360 00:17:35,757 --> 00:17:36,057 right? 361 00:17:36,282 --> 00:17:39,162 Well, I, I thought, uh, I'm, I'm, I'm gonna play dumb here. 362 00:17:39,192 --> 00:17:41,592 I thought we just talked about that, the charge 363 00:17:41,682 --> 00:17:43,842 No, that's just to restore the data. 364 00:17:43,842 --> 00:17:45,642 But to actually transfer the data 365 00:17:46,152 --> 00:17:46,512 Right, 366 00:17:46,692 --> 00:17:47,802 is a cost as well. 367 00:17:47,832 --> 00:17:48,432 So for 368 00:17:48,432 --> 00:17:53,472 instance, if your infrastructure exists on premises to pull your data from 369 00:17:53,502 --> 00:17:58,002 the cloud back to your data center, they're going to typically charge you. 370 00:17:58,992 --> 00:17:59,382 Yeah. 371 00:18:00,132 --> 00:18:01,542 Which you now need to factor in. 372 00:18:01,542 --> 00:18:05,052 And once again, this goes all the way back to the active archive definition, right? 373 00:18:05,382 --> 00:18:09,852 You're reading this data at some frequency, right? 374 00:18:09,852 --> 00:18:12,672 It's not just staying up there and you're never getting it back, right? 375 00:18:12,672 --> 00:18:18,252 So in addition to paying the restore cost, right, you also have to pay 376 00:18:18,252 --> 00:18:22,572 the data, transfer out cost or egress cost, and that could add up, 377 00:18:22,572 --> 00:18:24,522 especially if your data set is large. 378 00:18:25,812 --> 00:18:32,832 Yeah, that, that's a, and and just to sort of put in, know, simpler language, 379 00:18:32,832 --> 00:18:35,142 those last two things you're paying. 380 00:18:35,442 --> 00:18:40,182 I, I, if you're using infrequently accessed object storage, and, and 381 00:18:40,182 --> 00:18:44,412 that's part of the CO and that's the, like Glacier Deep Archive, for example, 382 00:18:44,412 --> 00:18:49,152 is one of these things where you, you get a really cheap monthly cost, but 383 00:18:49,152 --> 00:18:52,512 that's based on the idea that you're never gonna bring any of it back. 384 00:18:52,812 --> 00:18:53,142 Right. 385 00:18:53,322 --> 00:18:57,102 But if you do bring it back, then they charge you extra just 386 00:18:57,102 --> 00:18:58,332 to take it out of the archive. 387 00:18:58,332 --> 00:18:58,992 And then you're right. 388 00:18:59,322 --> 00:19:04,092 There is then the cost to, to bring it back so that first cost would happen, 389 00:19:04,092 --> 00:19:05,502 whether it leaves the cloud or not. 390 00:19:05,502 --> 00:19:09,132 Just like if, if they transferred it within Google, I'm sorry. 391 00:19:09,132 --> 00:19:13,842 Or within AWS, uh, and then that second cost is when you bring 392 00:19:13,842 --> 00:19:15,072 it back to your environment. 393 00:19:15,072 --> 00:19:17,562 Yeah, that's, these are really good, these are really good points. 394 00:19:18,192 --> 00:19:23,232 Um, the other, and, and, and just really quickly to talk about 395 00:19:23,232 --> 00:19:24,432 the, the data integrity issue. 396 00:19:25,542 --> 00:19:28,362 Tape is better at writing ones and zeros than disc. 397 00:19:28,782 --> 00:19:34,722 Um, and, uh, fi, um, um, flash, right? 398 00:19:35,352 --> 00:19:35,562 Why? 399 00:19:35,562 --> 00:19:36,702 What, what's with, what's with the 400 00:19:36,822 --> 00:19:37,902 no, I keep, 401 00:19:38,112 --> 00:19:38,832 WTF? 402 00:19:38,832 --> 00:19:39,367 Why the face? 403 00:19:39,942 --> 00:19:45,132 So I'm, I'm thinking though, so I agree that it is better 404 00:19:45,612 --> 00:19:46,932 There's no agree or disagree. 405 00:19:46,932 --> 00:19:48,462 It's just a scientific fact. 406 00:19:48,462 --> 00:19:48,582 Facts. 407 00:19:48,822 --> 00:19:52,562 but I do wonder with object storage 408 00:19:53,102 --> 00:19:59,552 and how they do sharding and everything else, if their error rates are actually 409 00:20:00,242 --> 00:20:04,802 better than single disc, maybe not quite approaching single tape level. 410 00:20:07,232 --> 00:20:08,042 So let, so let, 411 00:20:08,132 --> 00:20:08,582 of 412 00:20:08,642 --> 00:20:09,062 that's it. 413 00:20:09,452 --> 00:20:10,172 It's a good point. 414 00:20:10,172 --> 00:20:12,002 I don't mean to cut you out, but that's a good point. 415 00:20:13,022 --> 00:20:17,702 But I don't think so because when the. 416 00:20:18,152 --> 00:20:24,032 When the, um, initial corruption happens with a single bit, 417 00:20:25,262 --> 00:20:28,892 they've, they, they can possibly. 418 00:20:29,282 --> 00:20:32,822 The point is that when they read the data back, they might not notice 419 00:20:32,822 --> 00:20:33,782 that the corruption has happened. 420 00:20:34,682 --> 00:20:38,402 It's just there, there are all kinds of error correction that you can 421 00:20:38,402 --> 00:20:45,032 do on top, but again, it's just a that, that you can try again, try 422 00:20:45,332 --> 00:20:48,032 to mitigate some of these problems. 423 00:20:48,032 --> 00:20:51,332 But it's a simple fact that disc wakes makes. 424 00:20:51,622 --> 00:20:57,082 Way more errors when writing ones and zeros to magnetic media than tape does. 425 00:20:57,412 --> 00:21:01,252 It's literally two orders of magnitude better from the best 426 00:21:01,252 --> 00:21:03,952 disc to the worst tape, right? 427 00:21:04,132 --> 00:21:07,312 And then the nicer tapes is another one, or even two orders of 428 00:21:07,727 --> 00:21:11,267 I agree that it's not gonna be the same. 429 00:21:11,627 --> 00:21:16,487 I'm just wondering if it's better when we talk about object storage, right? 430 00:21:16,487 --> 00:21:21,377 Or a system like AWS versus like your traditional disc, right? 431 00:21:21,647 --> 00:21:26,717 Because of the additional protections that some of these storage vendors, uh, create. 432 00:21:27,512 --> 00:21:29,582 I'm not gonna say it's gonna beat tape, 433 00:21:30,032 --> 00:21:34,772 but I do wonder if it, maybe the delta is just a little bit better. 434 00:21:36,422 --> 00:21:38,042 I follow what you're saying. 435 00:21:38,792 --> 00:21:41,432 I'm just, I'm just saying. 436 00:21:41,492 --> 00:21:42,452 Well, so here's the thing. 437 00:21:43,082 --> 00:21:48,302 People write to tape with object storage too, and we write to disc 438 00:21:48,302 --> 00:21:50,582 with object storage too, right? 439 00:21:50,852 --> 00:21:57,062 So your, if, if we've got object storage on both sides, then it's 440 00:21:57,062 --> 00:21:58,832 still better on tape than disc. 441 00:21:59,362 --> 00:22:00,392 Yes, that is true. 442 00:22:01,067 --> 00:22:01,487 Right. 443 00:22:01,697 --> 00:22:03,917 I mean, we could argue, we could, we could, we'd have to do a 444 00:22:03,917 --> 00:22:08,807 lot of research to figure out whether, how much better it is. 445 00:22:08,807 --> 00:22:11,087 But the, the simple fact is it's better. 446 00:22:11,177 --> 00:22:11,537 Right. 447 00:22:11,537 --> 00:22:13,547 By the way, I, I like you challenging me on this, but, 448 00:22:13,547 --> 00:22:15,527 but that's just, that's just it. 449 00:22:15,767 --> 00:22:18,167 And then let's talk about the other thing. 450 00:22:18,167 --> 00:22:24,317 And again, you could argue that maybe, um, object storage would notice this. 451 00:22:25,742 --> 00:22:28,772 And, it may notice it, but it might not. 452 00:22:28,832 --> 00:22:31,632 And that is long term, a bit rot. 453 00:22:32,232 --> 00:22:35,232 Um, bit rot is way worse on this than it is on tape. 454 00:22:35,412 --> 00:22:35,802 Yes. 455 00:22:35,802 --> 00:22:36,312 Possibly. 456 00:22:37,282 --> 00:22:37,502 And, 457 00:22:37,947 --> 00:22:42,417 of the, uh, object storage stuff would maybe notice that the bit rot 458 00:22:42,417 --> 00:22:45,237 happened, uh, and then self-repair. 459 00:22:45,387 --> 00:22:46,557 I would hope so. 460 00:22:46,947 --> 00:22:54,387 Um, but my, but the alternative, the common belief is that tape is bad 461 00:22:54,387 --> 00:22:58,227 for long-term storage and tape is bad at writing data in the first place. 462 00:22:58,227 --> 00:22:59,547 And I'm saying it's actually better, 463 00:23:00,087 --> 00:23:01,947 No, I, I, I don't disagree with that at all. 464 00:23:02,277 --> 00:23:06,237 Uh, the other thing to also remember is cloud isn't something magical. 465 00:23:06,237 --> 00:23:08,157 It's just someone else's server, 466 00:23:08,842 --> 00:23:09,042 Yeah. 467 00:23:09,162 --> 00:23:09,252 It's, 468 00:23:09,717 --> 00:23:10,077 right? 469 00:23:10,872 --> 00:23:12,432 it's not a, it is not magical. 470 00:23:12,852 --> 00:23:15,882 Um, but let's talk about, let's talk about his two, his 471 00:23:16,137 --> 00:23:16,527 Sorry. 472 00:23:16,527 --> 00:23:17,127 I, sorry. 473 00:23:17,127 --> 00:23:19,587 that's, that's, those are the thoughts that ran through my head 474 00:23:19,857 --> 00:23:22,017 when you said cloud versus tape, and I was like, huh. 475 00:23:22,962 --> 00:23:24,582 So he, he, he has two claims. 476 00:23:24,582 --> 00:23:28,212 One is, he's saying that, that it takes too long to, if, if you're 477 00:23:28,212 --> 00:23:30,972 doing an active archive kind of thing, it takes too long to load. 478 00:23:31,452 --> 00:23:36,432 And second, it's, um, that, you know, you, you, you might, it might not load. 479 00:23:36,852 --> 00:23:37,302 Okay. 480 00:23:37,812 --> 00:23:38,412 Um, 481 00:23:39,132 --> 00:23:39,372 Does 482 00:23:39,372 --> 00:23:40,122 he talk about? 483 00:23:40,332 --> 00:23:40,542 Yeah. 484 00:23:40,542 --> 00:23:42,402 What does he mean by takes too long? 485 00:23:42,462 --> 00:23:42,762 Okay. 486 00:23:42,822 --> 00:23:43,092 Oh, 487 00:23:43,302 --> 00:23:46,152 Well, let's do the second one first, because I think it's just total bunk. 488 00:23:46,542 --> 00:23:48,402 I, I, I said, what are you talking about? 489 00:23:48,642 --> 00:23:48,972 Right. 490 00:23:49,062 --> 00:23:51,217 Maybe if we're talking about LTO from. 491 00:23:51,647 --> 00:23:51,867 Or. 492 00:23:52,077 --> 00:23:53,007 10, 15 years 493 00:23:53,037 --> 00:23:57,087 Or it's that one tape drive that you had at that one company you worked for like 494 00:23:57,117 --> 00:24:00,927 25 years ago where you would write it, you would put it out and you'd put it 495 00:24:00,927 --> 00:24:02,367 in a different one and it wouldn't work. 496 00:24:02,877 --> 00:24:03,207 Yeah. 497 00:24:03,207 --> 00:24:03,867 Yeah. 498 00:24:03,867 --> 00:24:05,457 that, was like 25, 30 years ago. 499 00:24:05,637 --> 00:24:07,467 We, we, yeah, we've worked on that. 500 00:24:07,467 --> 00:24:07,827 Right? 501 00:24:07,947 --> 00:24:12,627 And there was, when LTO was early, there was this concept of a loss leader, right. 502 00:24:12,627 --> 00:24:15,927 Where, you know, but it, it was fixable, right? 503 00:24:15,927 --> 00:24:18,057 It wasn't, your data wasn't in question. 504 00:24:18,057 --> 00:24:20,247 It was fixable and it was also rare. 505 00:24:20,727 --> 00:24:24,507 Uh, whereas when something bad happens to a disc, it's not fixable. 506 00:24:24,717 --> 00:24:25,947 You can, you just swap it out. 507 00:24:26,067 --> 00:24:26,427 Right. 508 00:24:27,867 --> 00:24:32,007 So this, this idea that it might not load, I was like, come on man. 509 00:24:32,157 --> 00:24:33,327 Like, what are you talking about? 510 00:24:33,327 --> 00:24:37,137 That, that's just, that's just a fud that, that has no, as many 511 00:24:37,137 --> 00:24:39,027 tapes as I have had in my life. 512 00:24:40,002 --> 00:24:46,392 I unless we talk, unless we go back to the early, early, early days, I, I don't have 513 00:24:46,392 --> 00:24:50,712 any memory of a tape that just wouldn't load that, that, that I couldn't fix. 514 00:24:51,342 --> 00:24:56,172 Um, so, but the, but the second one is, you know, it's, it's both 515 00:24:56,172 --> 00:25:02,442 a valid and an invalid claim and that is that disc will take longer. 516 00:25:02,592 --> 00:25:06,402 I'm sorry, tape will take longer to load than disc. 517 00:25:07,242 --> 00:25:07,662 Um. 518 00:25:08,772 --> 00:25:09,462 So it's both. 519 00:25:09,462 --> 00:25:13,212 It's it's valid and it is, and it's invalid for multiple reasons. 520 00:25:13,242 --> 00:25:20,922 One that I think even in an active archive situation, the idea is this stuff got 521 00:25:20,922 --> 00:25:22,722 archived and now you're bringing it back. 522 00:25:22,962 --> 00:25:25,452 The average load time of a tape is around two and a half 523 00:25:25,452 --> 00:25:26,682 minutes last time I checked. 524 00:25:26,682 --> 00:25:31,002 And so if you've, if you've archived it, that means you 525 00:25:31,002 --> 00:25:33,222 haven't looked at it in Right. 526 00:25:33,447 --> 00:25:34,887 The expectation is very low. 527 00:25:35,097 --> 00:25:36,357 expectation is very low. 528 00:25:36,357 --> 00:25:39,237 If it comes back in two and a half minutes, what? 529 00:25:39,237 --> 00:25:40,347 What's the problem? 530 00:25:40,527 --> 00:25:41,007 Right? 531 00:25:41,547 --> 00:25:42,237 Deep archive. 532 00:25:42,237 --> 00:25:43,527 I think it's 12 hours, 533 00:25:44,277 --> 00:25:44,577 right? 534 00:25:44,577 --> 00:25:45,297 For normal 535 00:25:45,297 --> 00:25:46,407 restore time 536 00:25:46,677 --> 00:25:49,197 you can, you can pay extra to get it back faster, but 537 00:25:49,647 --> 00:25:50,937 and by the way. 538 00:25:52,242 --> 00:25:54,762 You know, I can't prove this because I don't work there. 539 00:25:54,762 --> 00:25:58,422 And if I did work there, I probably wouldn't be able to tell you the, 540 00:25:58,422 --> 00:26:02,082 when they designed glacier and you know, and Glacier Deep Archive, they 541 00:26:02,082 --> 00:26:07,992 designed that SLA with tape in mind, right there is, why is it 12 hours? 542 00:26:07,992 --> 00:26:10,272 If it's disc, why is it any kind of hours? 543 00:26:10,272 --> 00:26:11,952 If it's disc, it's clearly tape. 544 00:26:12,012 --> 00:26:16,122 And by the way, that means it's tape on a shelf, meaning 545 00:26:16,122 --> 00:26:18,072 it's not tape in a tape robot. 546 00:26:19,092 --> 00:26:21,462 Right, because why would it need to be 12 hours? 547 00:26:21,462 --> 00:26:24,792 I, maybe, maybe it's there, there's, you know, it's, if it's 548 00:26:24,792 --> 00:26:28,812 a high volume library, there's a, a buildup of requests for tapes. 549 00:26:28,812 --> 00:26:30,972 But, um, 550 00:26:31,377 --> 00:26:33,177 But we don't know exactly what they're doing. 551 00:26:33,177 --> 00:26:33,387 This 552 00:26:33,387 --> 00:26:33,477 is 553 00:26:33,642 --> 00:26:34,872 don't know exactly what they're doing. 554 00:26:34,872 --> 00:26:35,262 Right. 555 00:26:35,682 --> 00:26:41,232 But I, I just, I don't think it's a valid complaint to say I don't, 556 00:26:41,982 --> 00:26:45,072 it's gonna be two to three minutes to get my file back, the file that 557 00:26:45,072 --> 00:26:47,292 I haven't looked at in six months. 558 00:26:48,267 --> 00:26:49,402 I just don't think that's a valid 559 00:26:49,682 --> 00:26:49,922 complaint. 560 00:26:50,607 --> 00:26:54,807 The expectation is that like the're, even in an active archive. 561 00:26:56,172 --> 00:26:56,652 If. 562 00:26:57,237 --> 00:27:00,627 If you're gonna put it on tape or in the cloud, which may be on tape, 563 00:27:01,347 --> 00:27:05,067 that is, that there's a, there's an understanding there that, that you're 564 00:27:05,067 --> 00:27:08,637 putting it in a place that's gonna require some effort to get it back. 565 00:27:09,027 --> 00:27:12,657 You're not gonna do that for stuff that you, you might need right away. 566 00:27:12,807 --> 00:27:14,637 You're doing, you're gonna put stuff there. 567 00:27:15,357 --> 00:27:18,717 Even if it's an active archive, even if it's a full HSM system, you're 568 00:27:18,717 --> 00:27:22,947 not gonna put stuff on tape or cloud that you might need tomorrow. 569 00:27:23,607 --> 00:27:26,187 This is a, I might need this in six months. 570 00:27:26,637 --> 00:27:29,907 And, and again, I just think, you know, if you've waited six months to 571 00:27:29,907 --> 00:27:33,057 pull the file out, if it comes out in three minutes from now, I, I don't, 572 00:27:33,057 --> 00:27:34,407 I don't see how that's a problem. 573 00:27:35,337 --> 00:27:38,997 Um, and if it is a problem, then I would suggest that you've got a bad design 574 00:27:39,702 --> 00:27:40,122 You probably 575 00:27:40,122 --> 00:27:42,102 don't wanna be actively archiving data. 576 00:27:42,762 --> 00:27:43,782 Yeah, yeah. 577 00:27:43,992 --> 00:27:47,322 You, you, well, you need a, you need a three-tiered archive. 578 00:27:47,322 --> 00:27:49,872 Then you need a, you need a disc archive. 579 00:27:49,872 --> 00:27:55,212 You need a, whether it's cloud or on-prem disc, you need a dis cache 580 00:27:55,542 --> 00:27:56,082 to your. 581 00:27:56,892 --> 00:27:59,592 You know, truly old stuff part, right? 582 00:28:01,092 --> 00:28:05,232 Um, so that's why, you know, when I read this thing, I, I, I didn't get 583 00:28:05,232 --> 00:28:06,812 hot and bothered, but I hearkened. 584 00:28:06,812 --> 00:28:08,282 Back to the tape Sucks. 585 00:28:08,282 --> 00:28:08,852 Move on. 586 00:28:08,852 --> 00:28:11,612 Bumper sticker from your former employer. 587 00:28:12,542 --> 00:28:13,232 Do you have it? 588 00:28:13,499 --> 00:28:15,629 I don't quite have that one, but I do. 589 00:28:15,629 --> 00:28:17,549 Oh wait, I have. 590 00:28:18,464 --> 00:28:18,974 What is that? 591 00:28:19,904 --> 00:28:20,414 What is that? 592 00:28:20,444 --> 00:28:22,934 Oh, is that, that is not cool. 593 00:28:26,084 --> 00:28:28,844 It says it was a truck that said lost tapes on the side. 594 00:28:29,099 --> 00:28:29,759 And then on the 595 00:28:29,759 --> 00:28:31,109 back it does say 596 00:28:33,169 --> 00:28:37,094 where, where is it a bumper sticker that says, tape sucks and move on. 597 00:28:37,724 --> 00:28:38,174 Yeah. 598 00:28:38,834 --> 00:28:41,684 I, I mean, you could argue that that was a very successful. 599 00:28:41,849 --> 00:28:42,299 Marketing 600 00:28:42,554 --> 00:28:43,664 Marketing campaign. 601 00:28:43,694 --> 00:28:45,614 'cause we still remember it all these years later. 602 00:28:45,614 --> 00:28:49,364 But I remember it because I thought it was bs, right? 603 00:28:49,424 --> 00:28:51,014 I mean, tape did suck. 604 00:28:51,014 --> 00:28:53,714 Tape does suck in certain ways, but 605 00:28:54,179 --> 00:28:54,929 if you use it in the, 606 00:28:54,929 --> 00:28:56,459 wrong way, it will suck. 607 00:28:56,579 --> 00:28:59,729 Just like anything else though, any other technology, use it in the wrong way 608 00:29:00,089 --> 00:29:03,449 or for an unintended use case and there will be issues. 609 00:29:04,784 --> 00:29:06,079 so if we talk about cloud. 610 00:29:07,289 --> 00:29:10,889 Really what it is, really what the discussion should be 611 00:29:10,919 --> 00:29:13,379 is on-prem versus off-prem. 612 00:29:14,039 --> 00:29:14,429 Right? 613 00:29:14,789 --> 00:29:16,349 Because cloud is tape. 614 00:29:17,069 --> 00:29:21,239 So if I'm putting the data in the cloud, especially older, 'cause you 615 00:29:21,239 --> 00:29:24,329 said it's talking about active archive, I'm probably putting it on tape. 616 00:29:24,329 --> 00:29:29,399 So it's not tape versus cloud, it's, it's on-prem tape versus off-prem, 617 00:29:29,429 --> 00:29:31,064 cloud, whatever that happens to be. 618 00:29:32,639 --> 00:29:39,509 is that the thing, or is it disc versus tape and on-prem versus off-prem? 619 00:29:42,419 --> 00:29:43,259 Sure. 620 00:29:43,559 --> 00:29:44,009 Right. 621 00:29:44,099 --> 00:29:52,019 I would just, I would just argue that like if you need the stuff you know, 622 00:29:52,049 --> 00:29:58,410 like that, then both tape and long and long term archive is not the place to do. 623 00:29:59,319 --> 00:29:59,609 Yeah. 624 00:30:00,014 --> 00:30:05,054 put it, if you, if you, if you, if you have access times measured in seconds, 625 00:30:05,534 --> 00:30:10,424 then you shouldn't be putting it on either tape or Glacier Deep Archive and the like. 626 00:30:10,499 --> 00:30:10,739 Yeah, 627 00:30:11,984 --> 00:30:12,434 Right. 628 00:30:12,479 --> 00:30:13,904 Are are we in agreement there? 629 00:30:14,189 --> 00:30:15,869 I am, but 630 00:30:15,914 --> 00:30:16,364 Yeah. 631 00:30:16,784 --> 00:30:17,474 What is with the, 632 00:30:17,894 --> 00:30:19,604 you always have a big butt, you know, that. 633 00:30:24,449 --> 00:30:30,029 By the way, that's the dirtiest line in, um, PeeWee's, Herman's big Adventure. 634 00:30:30,869 --> 00:30:34,709 She says, I know, but, and he goes, everybody I know has a big, but 635 00:30:37,049 --> 00:30:43,874 The, the reason I mentioned that is even if, so you said tape 636 00:30:43,874 --> 00:30:45,824 versus Glacier Deep Archive, right? 637 00:30:45,824 --> 00:30:46,994 Or whatever else in the cloud, 638 00:30:47,159 --> 00:30:48,359 it's some, something like that. 639 00:30:48,359 --> 00:30:48,629 Yeah. 640 00:30:49,304 --> 00:30:49,994 I think 641 00:30:52,064 --> 00:30:57,524 there are cases where you could still leverage. 642 00:30:58,949 --> 00:31:00,419 Oh, but that's still considered tape though. 643 00:31:01,274 --> 00:31:04,544 No, what I'm saying is the access time of Glacier Deep Archive 644 00:31:04,829 --> 00:31:05,669 Necessitates. 645 00:31:06,344 --> 00:31:08,144 in the sets it it is, is similar. 646 00:31:08,294 --> 00:31:09,464 It's actually worse than tape. 647 00:31:10,214 --> 00:31:10,634 Right? 648 00:31:11,204 --> 00:31:13,604 So if what he's saying is you get a better access time, well 649 00:31:13,604 --> 00:31:18,134 then that means he's using, like, he's not using any of the tiers. 650 00:31:18,944 --> 00:31:21,344 That have massive savings. 651 00:31:21,344 --> 00:31:23,054 So now we're talking huge difference 652 00:31:23,099 --> 00:31:26,159 It's S3 or, or it's like a normal 653 00:31:26,414 --> 00:31:28,004 S3 infrequent access. 654 00:31:28,034 --> 00:31:28,454 Right. 655 00:31:29,114 --> 00:31:29,864 Um, 656 00:31:30,029 --> 00:31:31,919 still very, very, very expensive. 657 00:31:32,444 --> 00:31:33,884 it's very expensive. 658 00:31:33,884 --> 00:31:34,334 Right. 659 00:31:34,694 --> 00:31:38,744 And then so then we're, so, so we're not talking about, I'm, I'm saying 660 00:31:38,744 --> 00:31:43,124 the closest thing to compare tape to is Glacier Deep Archive and the like. 661 00:31:43,484 --> 00:31:43,934 Right. 662 00:31:44,414 --> 00:31:47,834 Um, and I'm saying that if you need. 663 00:31:48,164 --> 00:31:52,896 Access time measured in seconds . You should not be putting it on tape or there, 664 00:31:53,076 --> 00:31:58,806 you should be putting it on, on-prem object storage or cloud object storage. 665 00:31:59,378 --> 00:32:00,849 this, this post got me all worked up. 666 00:32:01,563 --> 00:32:01,683 I 667 00:32:02,198 --> 00:32:05,498 just like, are you forgetful? 668 00:32:05,528 --> 00:32:08,348 Because literally three minutes ago you're like, oh yeah, no, 669 00:32:08,348 --> 00:32:09,963 this post didn't get me hot and. 670 00:32:13,778 --> 00:32:14,858 It got me all worked up. 671 00:32:15,008 --> 00:32:16,568 I was like, come on man. 672 00:32:16,628 --> 00:32:17,738 Come on, man. 673 00:32:18,458 --> 00:32:20,498 Doing my, doing my Biden impersonation. 674 00:32:20,498 --> 00:32:21,458 Come on man. 675 00:32:23,888 --> 00:32:30,278 Um, yeah, so it, it's about your requirements and if your requirements 676 00:32:30,278 --> 00:32:31,688 are measured in seconds, you're right. 677 00:32:31,688 --> 00:32:34,538 Tape's the wrong place, but so is Glacier Deep Archive, so 678 00:32:34,583 --> 00:32:37,973 But even if your time is measured in seconds, 679 00:32:38,468 --> 00:32:38,948 yeah. 680 00:32:38,993 --> 00:32:42,528 also make sure you understand where your data exists because. 681 00:32:44,378 --> 00:32:50,318 Even if you had used, say, infrequent access S3 with time measured in 682 00:32:50,318 --> 00:32:54,278 seconds, if you have to pull the data back on premises, you're 683 00:32:54,278 --> 00:32:55,178 gonna blow it out of the water. 684 00:32:55,178 --> 00:32:57,098 It doesn't matter, right? 685 00:32:57,578 --> 00:33:02,498 So the only time it makes sense to use cloud and S3, infrequent access for 686 00:33:02,498 --> 00:33:07,508 this particular use case, where you need seconds is if your source data or 687 00:33:07,508 --> 00:33:09,008 your compute also runs in the cloud. 688 00:33:11,093 --> 00:33:11,693 Yes. 689 00:33:13,313 --> 00:33:13,913 Yes. 690 00:33:14,033 --> 00:33:14,483 Good point. 691 00:33:14,573 --> 00:33:14,963 Good point. 692 00:33:14,963 --> 00:33:15,383 Good point. 693 00:33:15,383 --> 00:33:17,578 Yes, yes, yes, yes, yes, yes. 694 00:33:18,083 --> 00:33:19,253 You are so right, sir. 695 00:33:20,048 --> 00:33:20,408 thank you. 696 00:33:20,408 --> 00:33:21,308 Can I have another. 697 00:33:21,833 --> 00:33:22,178 my new, 698 00:33:29,028 --> 00:33:29,888 oh, the memories. 699 00:33:30,503 --> 00:33:31,283 Um. 700 00:33:32,273 --> 00:33:35,243 The shout out to my fellow shell backs. 701 00:33:36,413 --> 00:33:39,983 Somebody knows what that means, but most people won't. 702 00:33:40,163 --> 00:33:43,613 Anyway, I'm a, I'm a golden shellback. 703 00:33:43,733 --> 00:33:46,913 For those of you that know what that means, the rest of you 704 00:33:47,273 --> 00:33:49,583 feel free to Google it anyway. 705 00:33:49,733 --> 00:33:55,133 Um, it means that I was both hazed and I, and I hazed, um, 706 00:33:56,273 --> 00:33:58,643 the world's oldest hazing ritual. 707 00:33:59,723 --> 00:34:00,113 Um. 708 00:34:01,238 --> 00:34:06,543 Formerly sponsored by the US Navy, um, and now it's a, a, 709 00:34:06,743 --> 00:34:08,588 a shadow of his former self. 710 00:34:09,188 --> 00:34:11,558 'cause apparently now it's bad to beat people. 711 00:34:14,903 --> 00:34:16,403 Oh, Curtis. 712 00:34:16,808 --> 00:34:17,438 I kid. 713 00:34:17,438 --> 00:34:18,098 I kid. 714 00:34:18,518 --> 00:34:18,998 All right. 715 00:34:18,998 --> 00:34:23,528 Well, um, thanks for, thanks for having the chat. 716 00:34:23,558 --> 00:34:24,998 This was, this was fun. 717 00:34:25,718 --> 00:34:27,668 You know, it would've been more fun if the guy was here. 718 00:34:28,227 --> 00:34:31,347 To defend his position, but that's why we do the podcast so then we can make fun of 719 00:34:31,347 --> 00:34:33,387 people without them being on the podcast. 720 00:34:35,092 --> 00:34:35,382 Yeah. 721 00:34:37,222 --> 00:34:40,317 And, and then, and then he could, he could record his own podcast where he 722 00:34:40,317 --> 00:34:42,027 talks about us and how stupid we are. 723 00:34:42,957 --> 00:34:43,437 Anyway. 724 00:34:43,827 --> 00:34:44,187 All right. 725 00:34:44,217 --> 00:34:50,287 Well, uh, thanks to listeners you are, why we do this, that, and I dunno, Prasanna 726 00:34:50,307 --> 00:34:51,627 and I got nothing else better to do. 727 00:34:52,677 --> 00:34:54,537 Uh, that is a wrap. 728 00:34:57,726 --> 00:35:02,406 The backup wrap up is written, recorded, and produced by me w Curtis Preston. 729 00:35:02,976 --> 00:35:07,776 If you need backup or Dr. Consulting content generation or expert witness 730 00:35:07,776 --> 00:35:10,356 work, check out backup central.com. 731 00:35:11,076 --> 00:35:14,196 You can also find links from my O'Reilly Books on the same website. 732 00:35:14,856 --> 00:35:18,846 Remember, this is an independent podcast and any opinions that 733 00:35:18,846 --> 00:35:22,746 you hear are those of the speaker and not necessarily an employer. 734 00:35:23,676 --> 00:35:24,366 Thanks for listening.