1 00:00:00,000 --> 00:00:15,255 Outro Music 2 00:00:15,255 --> 00:00:15,995 Rinat: Hi everyone. 3 00:00:16,185 --> 00:00:21,905 Welcome to tech talk, a podcast where Amit and I talk about all things tech. 4 00:00:22,155 --> 00:00:23,525 We don't just talk about tech. 5 00:00:23,660 --> 00:00:28,320 We also talk about how the tech impacts our lives and the society altogether. 6 00:00:28,890 --> 00:00:31,270 This time we're going to talk about data centers. 7 00:00:31,670 --> 00:00:36,130 Data centers is one of those topics that , all of us have heard off. 8 00:00:36,170 --> 00:00:39,340 And even if you didn't, you could probably guess what data 9 00:00:39,340 --> 00:00:41,340 centers are anyway from its name. 10 00:00:41,890 --> 00:00:44,420 There's so many things I know about data centers. 11 00:00:44,510 --> 00:00:49,120 but as soon as I think about talking about something, I feel like all of 12 00:00:49,120 --> 00:00:54,900 those different knowledge of data centers comes together in an unorganized way. 13 00:00:55,330 --> 00:00:59,290 In this conversation, I'm looking to gather my thoughts, organize all 14 00:00:59,290 --> 00:01:03,242 the different knowledges I have and also learn something new, maybe so 15 00:01:03,272 --> 00:01:05,722 thank you for tuning in and let's. 16 00:01:05,917 --> 00:01:07,657 talk about data centers. 17 00:01:08,082 --> 00:01:10,212 Amit: Well, Thanks so much for that introduction. 18 00:01:10,542 --> 00:01:13,842 Data center is quite a relevant topic, especially during 19 00:01:13,842 --> 00:01:15,232 the age of cloud computing. 20 00:01:15,542 --> 00:01:18,962 But I think before we talk about data center, we need to actually talk about 21 00:01:19,222 --> 00:01:25,052 how People used to store data, transfer data, et cetera, before data centers. 22 00:01:25,102 --> 00:01:27,702 Companies used to manage their own IT infrastructure. 23 00:01:27,752 --> 00:01:31,142 And when I say IT infrastructure, that means all the computers that 24 00:01:31,142 --> 00:01:32,752 were provided in the offices. 25 00:01:33,042 --> 00:01:37,652 So suppose an office is a physical location at one physical location, 26 00:01:37,652 --> 00:01:39,032 you have say a hundred computers. 27 00:01:39,252 --> 00:01:43,202 You have to manage, update software make sure they're all secure, 28 00:01:43,252 --> 00:01:46,472 they're all connected and the backups are also in place. 29 00:01:46,792 --> 00:01:49,802 So that's how it all started from there. 30 00:01:50,192 --> 00:01:54,072 if you scale it up, so now you have say multiple locations and you 31 00:01:54,072 --> 00:01:58,402 still manage your computers, but each location has its own IT team. 32 00:01:59,027 --> 00:02:03,257 But what if you wanted to have a data center, say, in one city, you have 33 00:02:03,257 --> 00:02:06,957 two different office locations or five different office locations, and you want 34 00:02:06,957 --> 00:02:12,317 to manage all the IT infrastructure from a central location, and a data center can 35 00:02:12,337 --> 00:02:14,437 give you a clue about what it actually is. 36 00:02:14,647 --> 00:02:17,937 It is basically storing the data in a central location. 37 00:02:18,097 --> 00:02:19,497 That's why it's called a data center. 38 00:02:20,812 --> 00:02:24,802 Rinat: So as soon as I, think about this definition, this is what starts 39 00:02:24,802 --> 00:02:28,302 the alarm bells in my head because one of the major movements what 40 00:02:28,302 --> 00:02:32,782 we're trying to do is decentralize everything when it comes to technology. 41 00:02:32,782 --> 00:02:36,032 There is the saying that don't put all your eggs in one basket. 42 00:02:36,512 --> 00:02:41,892 From that perspective, data is one of the most important assets a company 43 00:02:41,892 --> 00:02:46,602 nowadays can have, whether you're an it company or even in non it company 44 00:02:47,072 --> 00:02:51,162 Putting all of your data in one place seems really counterintuitive. 45 00:02:51,172 --> 00:02:52,552 So what is the deal here? 46 00:02:52,562 --> 00:02:55,792 Amit: I don't know the exact answer, but if I take a guess, I think it 47 00:02:55,822 --> 00:03:00,202 is because you want to manage the infrastructure in a more scalable way. 48 00:03:00,252 --> 00:03:04,817 Because if you have to manage, say 100 computers, And the data 49 00:03:04,827 --> 00:03:07,777 on those computers, then it takes a lot of time and effort. 50 00:03:08,097 --> 00:03:14,077 Plus you have to go physically to separate locations restore the hard drives manage 51 00:03:14,077 --> 00:03:15,977 the network infrastructure, et cetera. 52 00:03:16,257 --> 00:03:19,127 But if you have a data center where all the backups are stored, 53 00:03:19,427 --> 00:03:23,217 then you know that, , in case a computer fails or in case a person 54 00:03:23,217 --> 00:03:26,927 leaves or whatever, I think the infrastructure is far easier to manage. 55 00:03:27,277 --> 00:03:29,637 Once we understand what are the different types of data center, I 56 00:03:29,677 --> 00:03:32,777 think that would make a lot of sense because I think a data center is 57 00:03:32,807 --> 00:03:34,707 not applicable for say a startup. 58 00:03:35,102 --> 00:03:37,382 It could be, but in a very different manner. 59 00:03:37,682 --> 00:03:40,762 Say you have computers and your IT infrastructure, your 60 00:03:40,762 --> 00:03:42,502 server in one central location. 61 00:03:42,712 --> 00:03:46,592 And when I say a server, the server is basically hosting a webpage, which 62 00:03:46,592 --> 00:03:47,952 can be accessed over the internet. 63 00:03:48,262 --> 00:03:50,932 So you're hosting your server where all the files are stored, your 64 00:03:50,932 --> 00:03:55,472 database is there, and people in the office are able to access the 65 00:03:55,472 --> 00:04:00,782 information on the database and work on the server and create your website. 66 00:04:01,372 --> 00:04:04,642 Plus you have your local machines and these desktop machines 67 00:04:04,662 --> 00:04:07,102 are managed by the IT team. 68 00:04:07,552 --> 00:04:08,552 So that's one part. 69 00:04:08,852 --> 00:04:12,942 Now, suppose you are a very big company, you have thousands of people and you 70 00:04:12,942 --> 00:04:17,162 want to manage the data of not just the employees, but also the product. 71 00:04:17,172 --> 00:04:20,652 So suppose you're making a CRM application and you want to manage that. 72 00:04:20,872 --> 00:04:25,062 those CRM application has a lot of data coming from the customers 73 00:04:25,062 --> 00:04:26,702 and you want to store that data. 74 00:04:27,162 --> 00:04:29,982 The data center is not the data of the people in the office, but 75 00:04:29,992 --> 00:04:32,422 the data is of the people who are actually using your service. 76 00:04:33,432 --> 00:04:37,102 I think the key thing is It's in the word data center. 77 00:04:37,102 --> 00:04:41,692 It's a center of data, but there are two types of data in any organization. 78 00:04:41,692 --> 00:04:46,342 One is the data of the employees who are working at the offices and the other 79 00:04:46,342 --> 00:04:50,922 data is the data of the customers who are accessing the services of the I. 80 00:04:50,972 --> 00:04:51,062 T. 81 00:04:51,062 --> 00:04:51,502 Company. 82 00:04:52,267 --> 00:04:54,067 So suppose I want to manage my own I. 83 00:04:54,107 --> 00:04:54,197 T. 84 00:04:54,197 --> 00:04:54,827 Infrastructure. 85 00:04:55,137 --> 00:05:00,967 So for that, I don't need to manage a lot of infrastructure in terms of data center. 86 00:05:01,327 --> 00:05:06,627 But if I want to manage my customers data, then I think that has to be managed 87 00:05:06,647 --> 00:05:10,307 in a far simpler way because I need to first make sure that the customer 88 00:05:10,307 --> 00:05:14,967 is able to access my infrastructure securely, safely and very quickly. 89 00:05:15,227 --> 00:05:17,047 So they get access to that information. 90 00:05:17,337 --> 00:05:20,717 Plus they can get access to that information 24 7. 91 00:05:20,987 --> 00:05:24,807 So it means it's always on plus it has to be secure so that no 92 00:05:24,807 --> 00:05:28,467 one is able to hack into that data and leak it onto the internet. 93 00:05:28,527 --> 00:05:31,497 So that's why it makes sense to have a a data center. 94 00:05:31,767 --> 00:05:35,087 This data center could be managed by the company. 95 00:05:35,147 --> 00:05:38,447 So suppose I build a CRM application, so I manage the data center. 96 00:05:38,767 --> 00:05:42,637 I take care of the physical space in which the servers are located. 97 00:05:42,897 --> 00:05:44,617 So there are two aspects to it, right? 98 00:05:44,807 --> 00:05:46,567 You have to manage the physical space. 99 00:05:46,597 --> 00:05:50,597 So people can't break in and the computers will get very hot so you 100 00:05:50,597 --> 00:05:53,757 have to provide some cooling, you have to maintain the humidity, you 101 00:05:53,757 --> 00:05:55,137 have to provide access control. 102 00:05:55,137 --> 00:05:57,487 So no unauthorized person can access the data. 103 00:05:57,807 --> 00:06:01,857 And you have to make sure that it's compliant with all the regulatory bodies. 104 00:06:02,207 --> 00:06:05,127 And then you have the physical infrastructure where in you have 105 00:06:05,127 --> 00:06:08,247 the computers, the networking equipment and the servers. 106 00:06:08,627 --> 00:06:10,137 So that's part of a data center. 107 00:06:10,237 --> 00:06:14,497 Now, in a different data center, the physical space itself is 108 00:06:14,507 --> 00:06:18,167 managed by a separate company, but the equipment, computer equipment 109 00:06:18,187 --> 00:06:20,217 is managed by the company itself. 110 00:06:20,817 --> 00:06:24,512 So suppose you are a company and I manage the physical space. 111 00:06:24,532 --> 00:06:28,992 You say, I want to rent your space a hundred square meters and I want 112 00:06:28,992 --> 00:06:31,452 to put my network servers there. 113 00:06:31,972 --> 00:06:33,202 I'll provide the security. 114 00:06:33,202 --> 00:06:36,312 I'll provide the cooling, but you bring your computer equipment. 115 00:06:36,542 --> 00:06:38,922 So that's co location data center. 116 00:06:39,422 --> 00:06:43,242 The third one is where I give this physical space. 117 00:06:43,357 --> 00:06:47,117 And I also provide the physical infrastructure of the computing equipment. 118 00:06:47,447 --> 00:06:50,757 So I'll provide the computers, I'll provide the space and then 119 00:06:50,757 --> 00:06:53,117 you can use that as a data center. 120 00:06:53,317 --> 00:06:57,457 And the final is cloud service provider like Amazon web services. 121 00:06:57,847 --> 00:06:59,907 Google cloud or Microsoft Azure. 122 00:07:00,307 --> 00:07:03,947 And these are infrastructure that is all virtualized, which can be 123 00:07:03,957 --> 00:07:05,617 accessed from anywhere in the world. 124 00:07:06,087 --> 00:07:08,397 So these are the four major data centers. 125 00:07:10,037 --> 00:07:11,987 Rinat: Wow, that's quite insightful. 126 00:07:11,987 --> 00:07:15,547 I actually didn't know that you can divide up different types 127 00:07:15,577 --> 00:07:17,107 of data centers in this way. 128 00:07:17,117 --> 00:07:19,687 It's a good thing to learn about as well. 129 00:07:20,597 --> 00:07:25,227 While you were describing all of these, I was also thinking that what are the 130 00:07:25,227 --> 00:07:30,907 motivation behind having each of these types and, going back to the first 131 00:07:31,087 --> 00:07:36,027 question I had is, why are we, increasing risk by having everything centralized? 132 00:07:36,027 --> 00:07:42,032 But then I thought it's all about isolating the problem and finding the 133 00:07:42,032 --> 00:07:46,002 solution for that particular problem, when someone says, don't put all of 134 00:07:46,002 --> 00:07:49,492 your eggs in one basket, they are trying to say that, okay, If you lose 135 00:07:49,492 --> 00:07:51,262 the basket, you lose all of your eggs. 136 00:07:51,792 --> 00:07:54,542 So how does it translate into data centers? 137 00:07:54,592 --> 00:07:57,802 If you put all of your data in one place, and then if the physical 138 00:07:57,802 --> 00:08:02,452 location is subject to an earthquake or a fire and you lose all of 139 00:08:02,452 --> 00:08:04,472 your data, so that's one problem. 140 00:08:04,782 --> 00:08:08,442 And there is a solution for that problem, which is having backup, 141 00:08:08,542 --> 00:08:12,792 online backup, offline backup, very quickly retrievable backup. 142 00:08:13,172 --> 00:08:18,262 And the same way there's all of these other problems that centralizing 143 00:08:18,282 --> 00:08:23,152 data comes with, there are isolated solutions for all of those as well. 144 00:08:23,322 --> 00:08:27,537 Ultimately, it's a addition and subtraction of what you gain from it. 145 00:08:27,547 --> 00:08:31,537 So you save a lot of cost by centralizing it. 146 00:08:31,997 --> 00:08:38,167 And even having two or three different backups, you would still save money by 147 00:08:38,177 --> 00:08:43,217 having a centralized location for all of your data because it would just reduce 148 00:08:43,887 --> 00:08:46,807 maintenance and management of it as well. 149 00:08:46,827 --> 00:08:50,677 So you would have to hire less people to manage it all together. 150 00:08:51,137 --> 00:08:53,127 And that's a lot of money as well. 151 00:08:53,497 --> 00:08:59,017 It's not just the infrastructure management, but also a huge team of 152 00:08:59,127 --> 00:09:03,867 solutions architects and data architects who would, make everything work together. 153 00:09:03,877 --> 00:09:10,527 After all of those calculation, it's more worth it to have a centralized data 154 00:09:10,527 --> 00:09:12,737 center, and that's why we have them. 155 00:09:13,237 --> 00:09:16,397 Amit: you say centralized data center, it doesn't mean that it's centralized 156 00:09:16,397 --> 00:09:18,587 as in it's just one data center, right? 157 00:09:18,787 --> 00:09:22,077 I think you have to look at the Cost and the growth of a company. 158 00:09:22,177 --> 00:09:26,777 So if a company is very small, you can manage one server at your own office. 159 00:09:26,897 --> 00:09:28,787 because you don't need a lot of hard disk space. 160 00:09:29,217 --> 00:09:30,747 You're not a Facebook or a Google. 161 00:09:30,967 --> 00:09:33,637 Imagine every like of a Facebook is stored. 162 00:09:34,372 --> 00:09:38,482 And if you liked a content 14 years ago, it's still stored. 163 00:09:38,832 --> 00:09:42,762 I graduated my university in 2006 and that's when Facebook came. 164 00:09:43,202 --> 00:09:44,532 It's 2024. 165 00:09:44,752 --> 00:09:46,792 So it's 18 years since I graduated. 166 00:09:47,132 --> 00:09:50,982 It has stored every single like every single photo, that I've 167 00:09:50,982 --> 00:09:52,932 clicked on every single comment. 168 00:09:53,332 --> 00:09:55,182 And there are 1 billion people accessing it. 169 00:09:55,192 --> 00:09:59,332 So imagine the amount of storage in terms of images, files, et cetera. 170 00:09:59,952 --> 00:10:04,332 Now, if you have to store that physically in one single location, of course, that 171 00:10:04,392 --> 00:10:09,312 can create a failure because if that place crashes, then all the data is lost. 172 00:10:09,532 --> 00:10:12,937 But when you create a data center, you have to look at what, The 173 00:10:12,997 --> 00:10:17,387 end goal is if I'm a very small company, I don't need a big data. 174 00:10:17,417 --> 00:10:18,787 I don't even need a data center. 175 00:10:18,787 --> 00:10:20,977 I can manage the infrastructure in my own company. 176 00:10:21,187 --> 00:10:25,357 I just need a small team to handle the maintenance of the servers to 177 00:10:25,387 --> 00:10:29,477 update the security patches to control the networking equipment and to have 178 00:10:29,517 --> 00:10:31,517 multiple hard disk in a RAID format. 179 00:10:31,767 --> 00:10:35,637 So RAID is basically to provide redundancies for hard disks. 180 00:10:35,807 --> 00:10:38,737 So basically if you have multiple hard disk, you can put them in 181 00:10:38,737 --> 00:10:42,857 different RAID configurations so that it can create automatic backups 182 00:10:42,857 --> 00:10:45,027 without you configuring it regularly. 183 00:10:45,387 --> 00:10:46,357 So that's one part. 184 00:10:46,597 --> 00:10:49,347 But suppose now you have A million customers. 185 00:10:49,637 --> 00:10:52,537 Now you can't have that server in your office because it 186 00:10:52,607 --> 00:10:53,927 will require a lot of space. 187 00:10:54,197 --> 00:10:57,607 plus you have to manage the cooling, then it's better you move to a separate 188 00:10:57,607 --> 00:10:59,877 physical location from your own office. 189 00:11:00,077 --> 00:11:03,917 So that way you can minimize the cost because now you have a location 190 00:11:03,937 --> 00:11:05,727 just for optimized for data center. 191 00:11:05,987 --> 00:11:07,527 you don't need tables. 192 00:11:07,527 --> 00:11:08,567 You don't need chairs. 193 00:11:08,767 --> 00:11:10,587 You just need a very big empty space. 194 00:11:10,847 --> 00:11:13,057 You need a good air conditioning. 195 00:11:13,067 --> 00:11:17,207 You need humidity control and you need people to guard and secure that location. 196 00:11:17,532 --> 00:11:21,132 So it's a very simple box in which you stack a lot of servers. 197 00:11:21,882 --> 00:11:26,702 If I want to now scale my infrastructure because I have that big empty space, 198 00:11:26,872 --> 00:11:28,412 I can scale it up very quickly. 199 00:11:28,862 --> 00:11:32,862 But if I have the servers in my own office, I need a different room. 200 00:11:33,177 --> 00:11:37,187 And then I have to figure out how to connect the servers into different rooms. 201 00:11:37,437 --> 00:11:41,487 But if it is just one physical location, it's just like attaching one cabinet, 202 00:11:41,737 --> 00:11:45,117 putting the hard drives, putting the network cable and then connecting it. 203 00:11:45,157 --> 00:11:46,857 And then it's all online, right? 204 00:11:48,317 --> 00:11:49,227 Rinat: Yeah, absolutely. 205 00:11:49,227 --> 00:11:52,637 You touched on a a very important aspect of I. 206 00:11:52,637 --> 00:11:52,897 T. 207 00:11:52,897 --> 00:11:57,437 Entrepreneurship and for our non techie listeners out there, you 208 00:11:57,437 --> 00:12:00,667 might think that if it's a technology company, then you should be able 209 00:12:00,667 --> 00:12:02,967 to handle having a data center. 210 00:12:03,087 --> 00:12:08,432 Obviously, depending on the need, the size and scale would differ uh, But, if you're 211 00:12:08,432 --> 00:12:11,042 an IT company, it's your bread and butter. 212 00:12:11,092 --> 00:12:15,772 And this is where I want to actually use an analogy of, a restaurant or a 213 00:12:16,152 --> 00:12:18,592 fast food, say, a burger restaurant. 214 00:12:19,012 --> 00:12:24,492 If it's just one branch and you're starting up, it makes more sense to 215 00:12:24,492 --> 00:12:31,322 prepare the burger patties and all of the individual items in your kitchen 216 00:12:31,452 --> 00:12:36,752 and then prepare and serve the gourmet burger and that's your business, but 217 00:12:36,762 --> 00:12:42,432 as your business grows and then you compare, start comparing it with fast 218 00:12:42,432 --> 00:12:48,242 food chains like McDonald's and the similar, where, you have grown to a degree 219 00:12:48,282 --> 00:12:53,912 that it no longer makes sense to create the burger patties at your location. 220 00:12:54,032 --> 00:12:58,862 You probably have a separate central location where you create the 221 00:12:58,862 --> 00:13:05,152 burger patties for five or ten or a hundred restaurants in the area. 222 00:13:05,432 --> 00:13:10,932 it also could come to a point where you outsource or delegate to another 223 00:13:10,932 --> 00:13:16,532 company who are better suited to cut and slice the potatoes into fries. 224 00:13:16,932 --> 00:13:21,322 Even though you are a fast food shop and, you don't think that just because 225 00:13:21,322 --> 00:13:25,162 of that McDonald's should, prepare all of their food by themselves. 226 00:13:25,202 --> 00:13:31,082 No, as you scale up the sort of the load of work becomes so high that it makes more 227 00:13:31,082 --> 00:13:37,582 sense to delegate or give that particular part of the task to a different company 228 00:13:37,582 --> 00:13:39,752 who would just do that and supply that. 229 00:13:39,802 --> 00:13:45,872 Even in IT company, you could specialize in say, for example, social media being 230 00:13:45,872 --> 00:13:47,842 a social media platform like Facebook. 231 00:13:48,442 --> 00:13:56,252 So your expertise is not to manage data in cloud, . So you might want 232 00:13:56,252 --> 00:13:59,632 to hire, I don't know what the actual configuration is Facebook's 233 00:13:59,642 --> 00:14:00,972 backend, but you might want to. 234 00:14:02,197 --> 00:14:06,757 Microsoft Azure or Amazon Web Services, who would handle all 235 00:14:06,757 --> 00:14:08,967 of your cloud infrastructure. 236 00:14:09,317 --> 00:14:12,597 And that way you can focus on what you do best. 237 00:14:12,697 --> 00:14:16,747 In case of Facebook, I don't know actually what they do best. 238 00:14:16,757 --> 00:14:22,427 But whichever IT company that we can talk about whatever their expertise is, they 239 00:14:22,427 --> 00:14:27,207 can focus on that and they can outsource or delegate this part of managing 240 00:14:27,207 --> 00:14:29,387 data which is a very important part. 241 00:14:29,397 --> 00:14:35,827 But, as it scales up, it becomes a very big challenge to handle by themselves. 242 00:14:36,207 --> 00:14:39,617 And that's where cloud data centers comes in. 243 00:14:39,657 --> 00:14:40,967 Amit: yeah, the analogy is right. 244 00:14:41,147 --> 00:14:44,727 The patties itself won't be manufactured at that single location. 245 00:14:45,157 --> 00:14:47,727 It's prepared at that location, but it's not manufactured. 246 00:14:47,727 --> 00:14:50,567 So patties are maybe prepared somewhere else. 247 00:14:50,567 --> 00:14:53,027 And then it's distributed to different locations. 248 00:14:53,037 --> 00:14:54,337 So yeah, that does make sense. 249 00:14:54,607 --> 00:14:59,017 I think with a data center, some companies can't even afford 250 00:14:59,027 --> 00:15:00,797 to have their own servers. 251 00:15:00,847 --> 00:15:05,647 So cloud service providers, provide access to a networking and 252 00:15:05,647 --> 00:15:07,617 storage solution over the internet. 253 00:15:08,217 --> 00:15:13,117 instead of buying multiple servers you basically rent it. 254 00:15:13,612 --> 00:15:19,012 And you say, I will buy a portion of your infrastructure with this processing 255 00:15:19,012 --> 00:15:21,742 power, this CPU, this storage etc. 256 00:15:22,232 --> 00:15:25,472 And I will rent it out for a certain period. 257 00:15:25,657 --> 00:15:27,917 And I will pay a subscription cost for it. 258 00:15:28,357 --> 00:15:31,667 They can provide the infrastructure plus the configuration. 259 00:15:31,687 --> 00:15:34,287 So they'll manage the firewall, they'll manage the DNS servers. 260 00:15:34,287 --> 00:15:35,927 You don't have to do the configuration. 261 00:15:36,387 --> 00:15:38,347 They will also install all the softwares. 262 00:15:38,577 --> 00:15:42,887 So you have the physical hardware, the hard disk RAM and the CPU. 263 00:15:43,177 --> 00:15:44,787 Then you have the configuration. 264 00:15:44,797 --> 00:15:46,457 You have to set up a firewall. 265 00:15:46,457 --> 00:15:47,517 You have to set up a DNS. 266 00:15:47,537 --> 00:15:50,517 You have to set up the server software and a database. 267 00:15:50,547 --> 00:15:51,307 So they do that. 268 00:15:51,497 --> 00:15:54,437 And then you have the software to manage all this, right? 269 00:15:54,647 --> 00:15:58,407 So there are different levels of what a cloud service provider can do. 270 00:15:58,597 --> 00:16:02,847 And based on your requirement, you can then say, okay, I want to pay for this 271 00:16:02,877 --> 00:16:04,677 and I will manage it in my own way. 272 00:16:04,677 --> 00:16:07,567 I'll configure it in my own way, but I'll rent it from them. 273 00:16:07,877 --> 00:16:10,617 And because I'm renting it from them, I don't have to worry 274 00:16:10,617 --> 00:16:12,187 about the physical security. 275 00:16:12,397 --> 00:16:13,627 I don't worry about cooling. 276 00:16:13,827 --> 00:16:15,937 I don't have to worry about paying the electricity bill. 277 00:16:16,177 --> 00:16:20,012 I don't have to worry about securing the physical location with say security 278 00:16:20,012 --> 00:16:23,242 staff, and then I don't have to worry about updating the infrastructure 279 00:16:23,242 --> 00:16:26,652 because I'm now getting the latest CPUs and the GPUs, which I need. 280 00:16:26,892 --> 00:16:29,132 So that's a benefit of a cloud service provider. 281 00:16:29,322 --> 00:16:34,292 Of course, as I mentioned, you have managed service providers, co location and 282 00:16:34,592 --> 00:16:36,902 companies having their own data center. 283 00:16:37,142 --> 00:16:40,102 So sometimes when you are such a big company, it makes sense 284 00:16:40,102 --> 00:16:41,522 to have your own data center. 285 00:16:42,337 --> 00:16:46,357 Amazon or Microsoft may be having their own data centers. 286 00:16:46,407 --> 00:16:49,487 Rinat: Microsoft are the best at one of the best in the market. 287 00:16:49,527 --> 00:16:54,567 So they would have their own, which is intranet and contained within 288 00:16:54,577 --> 00:16:59,827 their company, which manages or maintains all of the data inside. 289 00:17:00,157 --> 00:17:04,257 Amit: need to give you an example, like the first cloud computing platform started 290 00:17:04,257 --> 00:17:06,647 with AWS, Amazon Web Services, right? 291 00:17:06,817 --> 00:17:08,177 But before that, what did we have? 292 00:17:08,207 --> 00:17:09,457 We had virtual machines. 293 00:17:10,627 --> 00:17:15,987 So before that, what we used to have is we used to virtualize the hardware 294 00:17:15,987 --> 00:17:17,897 itself, virtualize on the hardware. 295 00:17:18,257 --> 00:17:22,707 So suppose you have one single machine with is very high capacity. 296 00:17:23,007 --> 00:17:25,947 What do you try to do is you try to say that, okay, I 297 00:17:25,947 --> 00:17:27,277 have this one single machine. 298 00:17:27,287 --> 00:17:30,807 Now I will install 10 different virtual machines on this. 299 00:17:30,847 --> 00:17:34,587 And this single hardware can serve 10 different people. 300 00:17:35,137 --> 00:17:36,257 So that's the virtual machine. 301 00:17:36,707 --> 00:17:41,217 Now, when Amazon became such a big company, it had to create a big data 302 00:17:41,217 --> 00:17:46,337 center to manage all the shipping orders and the logistics and the 303 00:17:46,557 --> 00:17:51,217 data that is actually created to create the whole website, right? 304 00:17:51,447 --> 00:17:54,797 So it has to manage all the images, it has to manage all the pricing, it has to 305 00:17:54,797 --> 00:17:56,847 manage all the authentication, login, etc. 306 00:17:57,087 --> 00:17:58,597 And plus store all the orders. 307 00:17:58,827 --> 00:18:01,617 And when it started growing big it created big data center. 308 00:18:01,737 --> 00:18:06,437 And then it thought, okay, I am managing all this data center and 309 00:18:06,447 --> 00:18:08,147 it's not getting fully utilized. 310 00:18:08,177 --> 00:18:11,037 Maybe I can rent a bit of it to other people. 311 00:18:11,647 --> 00:18:12,837 That's how cloud service started. 312 00:18:14,817 --> 00:18:17,827 Rinat: Yeah, that's such an interesting way of new product 313 00:18:17,857 --> 00:18:19,167 coming into market, right? 314 00:18:19,167 --> 00:18:23,347 A lot of tech entrepreneurs have built their whole business. 315 00:18:23,457 --> 00:18:25,347 They were initially doing something else. 316 00:18:25,357 --> 00:18:31,677 But in order to support that something else, they created even a better product. 317 00:18:32,107 --> 00:18:35,907 And I think some of our guests in the past were also talking 318 00:18:35,907 --> 00:18:37,937 about, similar kind of things. 319 00:18:37,937 --> 00:18:42,757 we had the Orkes for Netflix where they were trying to create something. 320 00:18:42,757 --> 00:18:43,097 And then. 321 00:18:43,232 --> 00:18:48,122 They ended up creating something which was really, necessary for a lot of people. 322 00:18:48,122 --> 00:18:49,362 So they ended up selling that. 323 00:18:49,802 --> 00:18:54,272 it's very interesting how you come up with new products and there 324 00:18:54,272 --> 00:18:57,592 are so many more opportunities coming every month, every year. 325 00:18:57,592 --> 00:18:58,882 People might think that. 326 00:18:58,937 --> 00:18:59,147 I. 327 00:18:59,147 --> 00:18:59,677 T. 328 00:18:59,737 --> 00:19:04,917 Has oversaturated with all of these new businesses, . But there is new 329 00:19:04,917 --> 00:19:08,197 inventions coming on every few days. 330 00:19:08,197 --> 00:19:11,187 And right now we're seeing a boom, with A. 331 00:19:11,547 --> 00:19:11,807 I. 332 00:19:12,287 --> 00:19:14,407 And new products will be made. 333 00:19:14,457 --> 00:19:17,657 Cloud services and all of these things. 334 00:19:17,677 --> 00:19:23,191 All of these were possible as the Internet became, more commonplace. 335 00:19:23,441 --> 00:19:29,501 That's why you could think of, okay, I have broadband, I have MBPS, not 336 00:19:29,541 --> 00:19:35,461 KBPS, so I could stream data, I could, collaborate online with multiple people. 337 00:19:35,741 --> 00:19:39,091 All of this became possible as the Internet speed grew. 338 00:19:39,531 --> 00:19:45,091 So now with AI and large language models and content generating 339 00:19:45,141 --> 00:19:48,951 models, which generates video, pictures, audio, all of these things. 340 00:19:49,341 --> 00:19:54,661 Now we're going to see, very soon , a new type of products that are emerging 341 00:19:54,661 --> 00:19:56,291 that people didn't even think about. 342 00:19:56,791 --> 00:20:01,591 It's an exciting place to watch the space and also think about yourself. 343 00:20:01,591 --> 00:20:06,971 Do something, maybe the thing that you set out to do that, that might not be 344 00:20:06,971 --> 00:20:08,661 the thing that makes you successful. 345 00:20:08,681 --> 00:20:11,551 But, in order to make that possible, you'll probably 346 00:20:11,551 --> 00:20:12,991 build something else around it. 347 00:20:13,091 --> 00:20:18,731 Which will then become the thing that makes you known and successful. 348 00:20:18,751 --> 00:20:21,371 I think one of the things that I always encourage everyone 349 00:20:21,451 --> 00:20:23,431 is to basically do something. 350 00:20:23,531 --> 00:20:27,051 implement something, take action on anything and you 351 00:20:27,051 --> 00:20:28,841 never know how it will turn out. 352 00:20:29,841 --> 00:20:31,841 Amit: This problem is faced by so many other people. 353 00:20:31,881 --> 00:20:34,191 So you can create a business out of it and monetize it. 354 00:20:34,511 --> 00:20:34,701 Yeah. 355 00:20:34,701 --> 00:20:35,871 I think you're absolutely right. 356 00:20:36,191 --> 00:20:39,641 We have talked about data center, but let's talk about the challenges of 357 00:20:39,721 --> 00:20:43,356 the data center and maybe what the future lies ahead for data centers. 358 00:20:43,616 --> 00:20:47,866 So one of the biggest challenge for opening any data center is not the 359 00:20:47,876 --> 00:20:52,976 hardware, not buying the software or the network switches, routers, etc. 360 00:20:52,976 --> 00:20:54,296 It's basically the energy. 361 00:20:54,956 --> 00:20:56,136 They consume a lot of energy. 362 00:20:56,596 --> 00:20:59,506 So you need to get subsidized electricity. 363 00:20:59,776 --> 00:21:07,186 Or subsidized energy in a form that is cheap for you to operate it plus because 364 00:21:07,206 --> 00:21:11,186 they consume a lot of energy to operate, they also generate a lot of heat. 365 00:21:11,196 --> 00:21:12,396 So you need a lot of cooling. 366 00:21:12,826 --> 00:21:18,026 So actually, data centers main cost comes from these particular things. 367 00:21:19,241 --> 00:21:23,051 So you have energy and you have the cooling and then you have to 368 00:21:23,051 --> 00:21:26,391 maintain the humidity because you don't want any computer infrastructure 369 00:21:26,391 --> 00:21:28,671 to get spoiled because of humidity. 370 00:21:29,231 --> 00:21:35,321 So you have to manage the energy, manage the humidity, manage the cooling. 371 00:21:35,641 --> 00:21:38,301 So once you've managed all that, you need to manage the physical 372 00:21:38,301 --> 00:21:41,991 location with security personnel, provide access control, et cetera. 373 00:21:42,261 --> 00:21:45,541 Plus you have to update the hardware because hardwares do fail because 374 00:21:45,551 --> 00:21:50,011 they are constantly running at a very extreme temperature in a very sensitive 375 00:21:50,011 --> 00:21:53,441 environment that in case one fails, you have to immediately replace them. 376 00:21:53,961 --> 00:21:57,661 Now, of course, all these servers have their own backups. 377 00:21:58,161 --> 00:22:01,631 In the same location or in different location in a different data centers. 378 00:22:01,961 --> 00:22:06,121 So if you think of AWS or Google, they don't have one physical location 379 00:22:06,131 --> 00:22:07,331 where all the data is stored. 380 00:22:07,581 --> 00:22:09,631 They have multiple data centers across the world. 381 00:22:09,921 --> 00:22:13,361 So they have one data center in say UK, one in Europe, 382 00:22:13,361 --> 00:22:14,991 one in US, one in Australia. 383 00:22:15,231 --> 00:22:16,896 So suppose you are based in Europe. 384 00:22:17,146 --> 00:22:20,906 They're going to serve the data from Europe servers, not from America, 385 00:22:20,936 --> 00:22:22,586 because the latency will be very high. 386 00:22:22,856 --> 00:22:25,776 That means that the time it takes for the data to travel 387 00:22:25,906 --> 00:22:27,886 from US to Europe is quite high. 388 00:22:28,096 --> 00:22:31,386 So they would rather serve it from a location closer to you so that 389 00:22:31,941 --> 00:22:35,471 They can instantly serve you the data, plus it creates a backup. 390 00:22:35,471 --> 00:22:39,081 So suppose there is a tsunami in Australia, then the data stored in 391 00:22:39,081 --> 00:22:41,401 Australia is also backed up in, say, U. 392 00:22:41,401 --> 00:22:41,551 S. 393 00:22:41,551 --> 00:22:42,031 and Europe. 394 00:22:42,271 --> 00:22:47,391 So not all the data is lost immediately because they have redundancies in place 395 00:22:47,421 --> 00:22:49,181 in terms of different physical location. 396 00:22:49,501 --> 00:22:50,391 So that's one part. 397 00:22:50,621 --> 00:22:54,621 So you have to manage the heating and to generate the energy The 398 00:22:54,621 --> 00:22:58,121 couple of option is you can go for generating your own energy. 399 00:22:58,121 --> 00:23:01,041 You can have solar cells but the solar panels will not 400 00:23:01,051 --> 00:23:02,611 generate enough electricity. 401 00:23:02,611 --> 00:23:07,101 So you have to maybe burn fossil fuels or you have to go for a nuclear power 402 00:23:07,101 --> 00:23:08,671 plant option, et cetera, et cetera. 403 00:23:08,871 --> 00:23:11,251 So it depends on how sustainable you want to go. 404 00:23:11,441 --> 00:23:14,791 And a lot of companies now they buy carbon credits and they say that, 405 00:23:14,791 --> 00:23:19,151 okay, our data center is completely carbon neutral because whatever, 406 00:23:19,201 --> 00:23:24,311 emission they produce, they are neutralizing it through some other means. 407 00:23:25,456 --> 00:23:29,926 Rinat: No, I was just going to add that it feels like quite an unethical thing to do. 408 00:23:29,926 --> 00:23:33,626 That's not the purpose of these things that are offered. 409 00:23:33,626 --> 00:23:37,176 Rather than actually working to reduce your carbon footprint, 410 00:23:37,186 --> 00:23:38,886 you're just buying it off. 411 00:23:38,976 --> 00:23:39,876 Amit: Think of it like this. 412 00:23:40,106 --> 00:23:44,236 Your main job is to provide IT services to multiple people or 413 00:23:44,236 --> 00:23:45,316 billions of people in the world. 414 00:23:45,336 --> 00:23:48,706 Your job is not to manage the carbon emissions. 415 00:23:49,101 --> 00:23:52,831 Someone else is managing that carbon emission for you, and they are saying 416 00:23:53,091 --> 00:23:57,801 that if you produce this much emission, we will plant this much trees so we 417 00:23:57,801 --> 00:24:00,071 will take care of that on your behalf. 418 00:24:00,141 --> 00:24:00,421 Rinat: Okay. 419 00:24:00,421 --> 00:24:02,201 I didn't realize that this is how it works. 420 00:24:02,201 --> 00:24:06,971 Of course, in that case, yeah, this is just another way of, doing the same thing. 421 00:24:06,971 --> 00:24:08,141 You're outsourcing 422 00:24:08,186 --> 00:24:10,326 Amit: Because that's not your main expertise, right? 423 00:24:10,576 --> 00:24:12,916 Rinat: That's actually good to know and understand. 424 00:24:12,996 --> 00:24:15,926 Amit: So once you have all this infrastructure, now you have to manage it. 425 00:24:15,946 --> 00:24:19,086 So someone will come, they will service the hard drives, they will 426 00:24:19,086 --> 00:24:22,646 update the softwares, they will, if there is any hard disk fail, or they 427 00:24:22,646 --> 00:24:25,556 have to retrieve some information because government has asked for it. 428 00:24:25,836 --> 00:24:28,126 So then there are different aspects to that as well. 429 00:24:28,206 --> 00:24:30,896 And then humans will have to enter the environment. 430 00:24:30,906 --> 00:24:34,836 So they'll have to make sure that They retrieve the right hard disk 431 00:24:34,866 --> 00:24:36,356 to retrieve the right information. 432 00:24:36,666 --> 00:24:39,896 We can replace that with robots so they can retrieve the data 433 00:24:39,916 --> 00:24:40,956 for you instead of humans. 434 00:24:41,226 --> 00:24:43,316 So that's one way where the data centers are heading. 435 00:24:43,656 --> 00:24:46,761 The other way is Using AI and machine learning. 436 00:24:46,761 --> 00:24:51,351 So you can use the algorithms to predict when the next hard disk would fail 437 00:24:51,431 --> 00:24:54,091 or when the next memory would fail. 438 00:24:54,091 --> 00:24:59,461 And you can based on that prediction, you can order the spare parts and be prepared 439 00:24:59,461 --> 00:25:01,331 to replace it as soon as it fails. 440 00:25:01,331 --> 00:25:06,391 So you have less downtime and you have instant connectivity. 441 00:25:06,671 --> 00:25:08,781 Then quantum computing is now being built. 442 00:25:08,791 --> 00:25:12,261 So how do you create a data center that has quantum computing 443 00:25:12,501 --> 00:25:14,131 provided to all your customers. 444 00:25:14,401 --> 00:25:15,901 So you have to think at that level. 445 00:25:16,271 --> 00:25:18,241 At a high level These are the challenges. 446 00:25:18,241 --> 00:25:21,081 And these are, this is the future in which the data center is heading. 447 00:25:21,291 --> 00:25:25,251 But I think one of the main things that you might have noticed, especially with 448 00:25:25,251 --> 00:25:30,451 apple intelligence getting launched recently the focus is AI running 449 00:25:30,481 --> 00:25:32,691 locally on your own hardware, right? 450 00:25:33,471 --> 00:25:35,631 So the data has to stay on your hardware. 451 00:25:35,771 --> 00:25:37,841 The algorithms have to run on your hardware. 452 00:25:38,221 --> 00:25:44,741 So in an age where data itself Is being managed on your edge device, 453 00:25:44,761 --> 00:25:49,001 which is called very close to your location or basically in your hand 454 00:25:49,311 --> 00:25:50,951 edge device could be your mobile phone. 455 00:25:50,951 --> 00:25:52,611 It could be a thermostat. 456 00:25:52,691 --> 00:25:54,361 It could be a doorbell. 457 00:25:54,551 --> 00:25:55,961 So these are edge devices, right? 458 00:25:56,831 --> 00:25:58,981 They're all connected to the internet, but they are actually 459 00:25:58,981 --> 00:26:00,591 closer to you than a data center. 460 00:26:01,091 --> 00:26:04,871 And the data is normally processed on board that computer and then 461 00:26:04,871 --> 00:26:07,151 stored at a different location, but it's processed there. 462 00:26:07,561 --> 00:26:10,321 So what happens to those edge devices? 463 00:26:10,321 --> 00:26:15,591 And because there are so many edge devices now or you can say internet of things, 464 00:26:15,621 --> 00:26:19,151 IOT devices, then where does that data go? 465 00:26:19,151 --> 00:26:21,681 You may not want to store all the information or all 466 00:26:21,711 --> 00:26:23,311 the logs in a data center. 467 00:26:23,311 --> 00:26:27,031 You might want to store it in your own house. 468 00:26:27,241 --> 00:26:31,891 So you might have your own backups or systems in place to do that, right? 469 00:26:31,981 --> 00:26:33,771 That information you don't want data center. 470 00:26:33,851 --> 00:26:34,131 Rinat: Absolutely. 471 00:26:34,161 --> 00:26:38,541 To implement AI decision making you need to have localized 472 00:26:38,591 --> 00:26:40,451 processing of all of those data. 473 00:26:40,471 --> 00:26:44,891 You don't want it to send it to a central location and be processed and then come 474 00:26:44,901 --> 00:26:46,491 back because that might be too slow. 475 00:26:46,881 --> 00:26:51,421 for this kind of IOT devices, you want it instantaneous as if you're talking 476 00:26:51,421 --> 00:26:56,571 to a robot helper in the kitchen you definitely want localized processing. 477 00:26:56,571 --> 00:26:58,611 So for that how would that be managed? 478 00:26:58,621 --> 00:27:04,641 That's whether, moving from Mbps to Gbps internet speed is a better 479 00:27:04,641 --> 00:27:07,191 option than localized processing. 480 00:27:07,191 --> 00:27:10,551 Because if you want to do localized, You have to think about the cost of 481 00:27:10,561 --> 00:27:15,991 hardware that needs to be present and that's the whole point of I. 482 00:27:15,991 --> 00:27:16,451 T. 483 00:27:16,481 --> 00:27:22,141 and, making everything remote that, make everything more efficient by having one 484 00:27:22,151 --> 00:27:28,671 thing in a central location, which does, which serves like 20 people or 20 devices. 485 00:27:29,141 --> 00:27:32,071 But if you need to do localized processing, that's 486 00:27:32,071 --> 00:27:33,501 a totally different story. 487 00:27:33,501 --> 00:27:34,861 Obviously it could go both way. 488 00:27:34,861 --> 00:27:36,861 We never know what the future holds. 489 00:27:36,891 --> 00:27:40,061 But, as you're saying that one of the options for data 490 00:27:40,131 --> 00:27:41,541 center is to move that way. 491 00:27:41,611 --> 00:27:48,171 I'm thinking that maybe it's more the kind of the overlords of the I. 492 00:27:48,171 --> 00:27:48,451 T. 493 00:27:48,991 --> 00:27:55,081 It might actually want to focus more on how to improve that communication so they 494 00:27:55,081 --> 00:27:57,511 can still do the processing remotely. 495 00:27:57,521 --> 00:28:03,511 The data transfer is so high speed that it feels instantaneous. 496 00:28:03,641 --> 00:28:06,451 Amit: I think The cloud data centers will still exist. 497 00:28:06,831 --> 00:28:09,951 What will happen is that a lot of processing will happen on board. 498 00:28:10,591 --> 00:28:12,791 So currently you fetch some information. 499 00:28:13,091 --> 00:28:17,071 you log into Amazon, you buy a product, you buy a laptop. 500 00:28:17,871 --> 00:28:21,851 All that information will be still stored on Amazon servers and you 501 00:28:21,851 --> 00:28:24,871 can retrieve the information from your mobile app from your laptop. 502 00:28:25,461 --> 00:28:30,171 But say you have and AI algorithm running on your smartphone and 503 00:28:30,181 --> 00:28:33,601 you want to edit some images that are stored on your smartphone. 504 00:28:33,631 --> 00:28:37,321 They are not stored on the, say Google drive or Google photos. 505 00:28:37,641 --> 00:28:39,281 I'm talking about an Android phone. 506 00:28:39,901 --> 00:28:41,101 So, that processing is local. 507 00:28:41,501 --> 00:28:42,861 Plus you have certain emails. 508 00:28:43,161 --> 00:28:46,881 So My emails are stored in a data center, but the AI 509 00:28:46,881 --> 00:28:48,801 processing is happening locally. 510 00:28:48,871 --> 00:28:50,431 So now there are two aspects to it. 511 00:28:50,701 --> 00:28:54,701 The data itself that is being retrieved from a data center and the processing. 512 00:28:54,931 --> 00:28:57,811 I don't want the processing to happen in a remote location. 513 00:28:57,811 --> 00:29:00,351 I want the processing to happen in this location. 514 00:29:00,361 --> 00:29:00,741 Why? 515 00:29:01,041 --> 00:29:06,081 Because the moment the AI algorithm works on my personal data, it 516 00:29:06,091 --> 00:29:10,341 trains itself and I don't want it to train on my personal data. 517 00:29:10,501 --> 00:29:14,091 I want it to be trained on public information, and this 518 00:29:14,091 --> 00:29:15,841 is private information, right? 519 00:29:17,051 --> 00:29:18,301 Rinat: Yeah I know. 520 00:29:18,301 --> 00:29:22,671 This is obviously a very important thing about privacy and, everyone 521 00:29:22,671 --> 00:29:26,441 should have the right to their own data and how they're used and everything, 522 00:29:26,861 --> 00:29:32,254 but at the same time, I also see the, Difficulty it brings in terms of I. 523 00:29:32,254 --> 00:29:32,554 T. 524 00:29:32,554 --> 00:29:34,904 And technological innovation, because when A. 525 00:29:34,934 --> 00:29:35,074 I. 526 00:29:35,074 --> 00:29:39,794 Is being trained is anonymized data, and once it learns the pattern, it has 527 00:29:39,794 --> 00:29:44,924 no understanding of individual personal data, but it has a overall whole 528 00:29:44,924 --> 00:29:49,284 understanding of all different kinds of email, all different kinds of people sent. 529 00:29:49,329 --> 00:29:51,659 Amit: but if you think about anonymized data, suppose you are 530 00:29:51,659 --> 00:29:53,309 writing an email to a company, right? 531 00:29:53,609 --> 00:29:56,389 And your personal company, and you will have names, you will sometimes 532 00:29:56,389 --> 00:29:59,769 have date of births, you will have some sensitive information in it, 533 00:29:59,769 --> 00:30:03,559 say banks have sensitive information, financial transactions, etc. 534 00:30:03,849 --> 00:30:07,479 And suppose you train and you use a AI in that environment, and 535 00:30:07,479 --> 00:30:09,279 now that AI model has learned. 536 00:30:09,429 --> 00:30:10,429 About all those things. 537 00:30:10,679 --> 00:30:13,739 And now someone who knows that this is the training data it has 538 00:30:13,739 --> 00:30:17,559 been trained on, you can actually figure out a way to retrieve that 539 00:30:17,559 --> 00:30:20,439 information from the algorithm, right? 540 00:30:20,709 --> 00:30:22,259 So you don't want that. 541 00:30:22,339 --> 00:30:26,899 So that's why when you talk to chat GPT, you don't say 542 00:30:26,899 --> 00:30:29,099 that, okay, this is my company. 543 00:30:29,129 --> 00:30:34,589 I work at say Amplify and I do this and this is my current problem. 544 00:30:34,589 --> 00:30:35,619 This is my boss name. 545 00:30:35,619 --> 00:30:36,739 I want to create an email. 546 00:30:36,959 --> 00:30:40,349 No, you don't want that information because that sensitive information, 547 00:30:40,559 --> 00:30:42,549 how do you think is getting anonymized? 548 00:30:42,589 --> 00:30:46,679 I don't think that if there are billions of people putting inputting 549 00:30:46,709 --> 00:30:49,489 all that information, do you think there are billions of people 550 00:30:49,699 --> 00:30:51,189 trying to anonymize the data? 551 00:30:51,419 --> 00:30:52,259 I don't think so. 552 00:30:52,704 --> 00:30:53,704 Rinat: No, absolutely not. 553 00:30:53,714 --> 00:30:55,514 That's actually a very good point. 554 00:30:55,534 --> 00:30:59,854 No, if those kind of information are stored then definitely, I 555 00:30:59,854 --> 00:31:03,234 was thinking about when you said localized processing for that, you 556 00:31:03,234 --> 00:31:05,999 need a processor powerful enough. 557 00:31:06,239 --> 00:31:10,329 If we're talking about IOT devices, then, adding each of these device 558 00:31:10,359 --> 00:31:15,739 with powerful enough processor to run AI algorithms is quite a costly 559 00:31:15,809 --> 00:31:19,429 Amit: So I think a lot of the mobile companies, they are now having different 560 00:31:19,429 --> 00:31:23,399 cores because I've been recently looking at thinking of upgrading my own laptop. 561 00:31:24,239 --> 00:31:25,689 So I was looking at different options. 562 00:31:25,769 --> 00:31:29,729 So they have different silicons and these silicons now have different cores. 563 00:31:29,839 --> 00:31:33,359 So some cores are just for cores which are optimized for computing tasks. 564 00:31:33,559 --> 00:31:35,929 And then there are some cores that are optimized for A. 565 00:31:35,929 --> 00:31:36,049 I. 566 00:31:36,049 --> 00:31:39,979 Tasks and a task means multiple threads, multiple processing, 567 00:31:39,989 --> 00:31:41,079 neural networks, etcetera. 568 00:31:41,079 --> 00:31:43,909 So they're optimized just for those types of tasks. 569 00:31:44,239 --> 00:31:47,919 And these silicons are now being built and they're getting installed 570 00:31:47,919 --> 00:31:50,589 on the smartphones or the laptops. 571 00:31:50,799 --> 00:31:53,767 So even in Intel series, you have the Intel Core series and 572 00:31:53,767 --> 00:31:55,347 the Intel core ultra series. 573 00:31:55,557 --> 00:32:00,137 The ultra series has AI cores in it and then the latest M1 chips from 574 00:32:00,397 --> 00:32:02,757 Apple, they have those chips as well. 575 00:32:03,017 --> 00:32:06,907 So the Apple intelligence won't work on say anything below Apple iPhone 576 00:32:06,967 --> 00:32:13,667 15 or any Mac book that doesn't have an M1 chip or a lower chip, right? 577 00:32:13,817 --> 00:32:16,697 So it will be limited by the hardware infrastructure. 578 00:32:16,957 --> 00:32:19,837 But over a period of time, that infrastructure cost will come down 579 00:32:20,117 --> 00:32:21,477 and it will become more accessible. 580 00:32:21,797 --> 00:32:27,177 And then more and more computation will be done on your devices. 581 00:32:27,197 --> 00:32:29,947 Even Google for their pixel devices, they manufacture their 582 00:32:29,947 --> 00:32:31,997 own tensor processing units. 583 00:32:32,277 --> 00:32:36,497 So I think there is definitely a possibility of Scaling it 584 00:32:37,657 --> 00:32:37,927 Rinat: Yeah. 585 00:32:37,927 --> 00:32:39,187 Future is exciting. 586 00:32:39,187 --> 00:32:42,617 There's so many groundbreaking innovations happened in the 587 00:32:42,617 --> 00:32:45,027 last one or two years with AI. 588 00:32:45,027 --> 00:32:48,947 And some of the things that are in the horizon, like quantum computing, 589 00:32:48,977 --> 00:32:51,227 which will turn everything upside down. 590 00:32:51,227 --> 00:32:55,507 And then obviously people, countries and companies are working 591 00:32:55,507 --> 00:32:58,117 towards getting AGI as well. 592 00:32:58,147 --> 00:33:03,167 And it's a very Exciting place to know about everything that's 593 00:33:03,167 --> 00:33:07,207 happening, and that's why our podcast here, I think, touches on all of 594 00:33:07,207 --> 00:33:10,277 these the relevant topics of time. 595 00:33:10,307 --> 00:33:14,667 And I hope our audience would continue listening to different topics that 596 00:33:14,667 --> 00:33:16,737 we talk about every few weeks. 597 00:33:16,837 --> 00:33:21,357 One of the things I have heard at one point is that Microsoft is thinking 598 00:33:21,357 --> 00:33:26,487 of putting their data centers on the seabed because it's so cold, it 599 00:33:26,497 --> 00:33:28,907 would just be naturally easier to 600 00:33:28,907 --> 00:33:32,097 Amit: or in Arctic locations or Antarctic locations where it's, 601 00:33:32,097 --> 00:33:33,497 but there we have permafrost. 602 00:33:33,747 --> 00:33:36,977 But then of course you have to bear in mind, how do you access it physically? 603 00:33:37,287 --> 00:33:39,287 Do you station staff there physically? 604 00:33:39,447 --> 00:33:41,047 What if the polar ice caps melt? 605 00:33:41,247 --> 00:33:46,047 What if there is a leak in the underwater data center, who fixes it? 606 00:33:46,097 --> 00:33:48,897 So then there are different kinds of challenges, but yes, of course, it 607 00:33:48,897 --> 00:33:52,762 solves one part of the problem, but then it creates other problems as well. 608 00:33:54,187 --> 00:33:55,007 Rinat: Absolutely. 609 00:33:55,007 --> 00:33:57,321 That is the story of all things. 610 00:33:57,761 --> 00:34:00,511 Amit: The main thing is that we have to now start thinking about 611 00:34:00,711 --> 00:34:02,021 space infrastructure as well. 612 00:34:02,341 --> 00:34:07,181 So once we start going to moon and Mars, we will then have different types of 613 00:34:07,181 --> 00:34:11,751 challenges because the data can't travel from moon to earth instantly, right? 614 00:34:11,881 --> 00:34:13,341 So there is a time lag. 615 00:34:13,541 --> 00:34:14,761 So you have to store it somewhere. 616 00:34:14,991 --> 00:34:18,681 Then you have to transmit it somewhere and then you have to provide good Connection. 617 00:34:18,921 --> 00:34:23,131 So how do you manage data centers, say, in a world where you have moon bases 618 00:34:23,151 --> 00:34:27,441 or your bases in Mars, so that will provide a different set of challenges. 619 00:34:27,721 --> 00:34:32,081 Now you have to provide communication between Earth, Moon, Mars, et cetera, 620 00:34:32,081 --> 00:34:36,891 Rinat: Yeah, the universal speed limit has just become too slow for us humans. 621 00:34:36,921 --> 00:34:43,451 Now, even just 100 years ago, we were just about measuring the speed of light. 622 00:34:43,541 --> 00:34:47,651 Amit: Now we're hitting this limit off the thing, but I think I was reading a 623 00:34:47,651 --> 00:34:52,721 blog recently where they were talking about how do you operate infrastructure 624 00:34:52,721 --> 00:34:56,721 in a very slow internet connection, right? 625 00:34:57,941 --> 00:35:02,101 What it means is suppose you don't have 100 megabyte connection, which we are. 626 00:35:02,101 --> 00:35:03,321 so used to right now, right? 627 00:35:03,601 --> 00:35:07,191 So what if you operate it at 10 mbps connection or what if you are 628 00:35:07,271 --> 00:35:11,261 operating in a One mbps connection or even 500 kbps connection. 629 00:35:11,671 --> 00:35:16,361 So that happens in Antarctic because they get internet only for a few hours a 630 00:35:16,361 --> 00:35:18,181 day when the satellite passes over them. 631 00:35:18,731 --> 00:35:20,631 And that's how they get the internet connection. 632 00:35:20,831 --> 00:35:21,921 And now it's very slow. 633 00:35:22,001 --> 00:35:26,941 So in that slow internet speed, what infrastructure can you operate and how 634 00:35:26,941 --> 00:35:28,911 can you manage that effect efficiently? 635 00:35:29,101 --> 00:35:31,441 I think that is another type of challenge, right? 636 00:35:31,621 --> 00:35:33,831 It's because we're so used to high speed internet. 637 00:35:34,011 --> 00:35:38,181 We think all problems can be solved with providing more bandwidth, etc. 638 00:35:38,351 --> 00:35:41,911 But, there are locations where you just cannot provide high speed. 639 00:35:42,506 --> 00:35:45,796 How do you manage infrastructure and how do you set expectations 640 00:35:46,296 --> 00:35:47,766 for that kind of environment? 641 00:35:48,186 --> 00:35:50,896 And I think that is again a beautiful challenge to solve. 642 00:35:52,311 --> 00:35:57,081 Rinat: let's touch base in terms of data center for our non techie folks out there. 643 00:35:57,121 --> 00:35:58,891 I hope you guys already learned a lot. 644 00:35:58,891 --> 00:35:59,561 I have. 645 00:36:00,191 --> 00:36:00,621 Thank you. 646 00:36:00,621 --> 00:36:04,551 Amit for all of these insights on data center and more. 647 00:36:04,921 --> 00:36:08,441 But just to bring it all in one one thought. 648 00:36:08,521 --> 00:36:13,071 Data center is in one or multiple physical location. 649 00:36:13,111 --> 00:36:15,671 We might think that it's virtual, but it's not. 650 00:36:16,071 --> 00:36:21,081 The same way when we talked about cloud in one of our previous episodes, we 651 00:36:21,141 --> 00:36:26,021 talked about cloud computing, cloud infrastructure, even though the name makes 652 00:36:26,021 --> 00:36:29,971 it sound like it's somewhere over there in the sky that we don't have to know 653 00:36:29,971 --> 00:36:32,131 about, or even it's, it may be virtual. 654 00:36:32,471 --> 00:36:37,771 But actually, It's quite closely related in this way that , even though we say 655 00:36:37,781 --> 00:36:42,071 it's in the cloud, it's probably in a data center somewhere, which is a 656 00:36:42,071 --> 00:36:47,681 physical location, heavily guarded, heavily, protected and environmentally 657 00:36:47,681 --> 00:36:52,571 controlled in a granular way to make sure that everything's functioning. 658 00:36:52,581 --> 00:36:56,751 this physical location could be anywhere as we've discussed today under the 659 00:36:57,946 --> 00:37:02,546 in arctics or in, in countries where electricity is probably cheaper. 660 00:37:03,186 --> 00:37:06,736 Many things to consider when you are thinking of data center, but 661 00:37:06,736 --> 00:37:12,246 the necessity of it is definitely something that is unavoidable for 662 00:37:12,246 --> 00:37:15,136 small companies and for large I. 663 00:37:15,136 --> 00:37:15,646 T. 664 00:37:15,766 --> 00:37:17,216 Organizations I. 665 00:37:17,216 --> 00:37:17,496 T. 666 00:37:17,496 --> 00:37:19,296 Or not even for banks. 667 00:37:19,326 --> 00:37:25,756 They have their own data centers a lot of the times for the very reason of security 668 00:37:25,756 --> 00:37:28,456 and privacy off of their customers. 669 00:37:28,846 --> 00:37:29,676 Whether you're an I. 670 00:37:29,676 --> 00:37:29,926 T. 671 00:37:29,926 --> 00:37:34,766 Company or not, as long as you're, a mid sized company in any industry, 672 00:37:34,766 --> 00:37:37,206 you would probably need data center. 673 00:37:37,216 --> 00:37:42,486 to be able to operate and provide services to your customer, and that's 674 00:37:42,486 --> 00:37:48,956 why data center is a very crucial bit of technology that is everywhere and is 675 00:37:48,956 --> 00:37:55,756 needed by everyone as anyone is using some sort of technology on a day to day life. 676 00:37:56,091 --> 00:38:02,041 Anyway, so it's a very sort of important topic to know about, learn about, 677 00:38:02,071 --> 00:38:07,621 and also to train yourself about, because as people are talking about, 678 00:38:07,631 --> 00:38:12,871 AI taking over people's jobs and which skills are becoming out of date, etc. 679 00:38:13,211 --> 00:38:16,271 Anything to do with data centers as it may evolve. 680 00:38:16,611 --> 00:38:22,221 The technology, but knowing about data center being trained and certified on 681 00:38:22,241 --> 00:38:28,161 various aspects of data center will always be in place because, AI or any other 682 00:38:28,161 --> 00:38:33,081 technology can't survive without data and it being transmitted very quickly. 683 00:38:33,641 --> 00:38:35,651 So that's why it's very important. 684 00:38:35,696 --> 00:38:38,676 Amit: I think one of the things that we are now seeing is that AI 685 00:38:38,696 --> 00:38:40,366 needs a lot of processing power. 686 00:38:40,646 --> 00:38:43,456 Most of the servers were built using CPUs, not GPUs. 687 00:38:43,636 --> 00:38:46,316 So now there are new data centers being built with GPUs. 688 00:38:46,536 --> 00:38:49,876 And the problem with GPUs is they consume far more energy than a CPU. 689 00:38:50,206 --> 00:38:52,756 So if you have a lot of GPUs here, hundreds and thousands 690 00:38:52,756 --> 00:38:53,496 of GPUs in one place. 691 00:38:53,806 --> 00:38:58,656 One physical location, it sometimes requires far more energy than is 692 00:38:58,656 --> 00:39:00,236 actually consumed in one city. 693 00:39:00,656 --> 00:39:06,636 It requires a very large amount of electricity and that electricity can be 694 00:39:06,636 --> 00:39:11,406 provided through maybe hydroelectric, nuclear, fossil fuels, et cetera. 695 00:39:11,676 --> 00:39:16,096 So it creates its own challenges now, especially with the infrastructure because 696 00:39:16,256 --> 00:39:20,076 you have to train these large language models with tons and tons of data. 697 00:39:21,851 --> 00:39:25,061 Rinat: We've talked about a lot of challenges around data center, and 698 00:39:25,201 --> 00:39:28,961 those are quite serious challenges that needs to be solved, even if 699 00:39:28,961 --> 00:39:32,741 we have a solution right now, there needs to be a better solution. 700 00:39:33,101 --> 00:39:37,301 So to our audience, can you guys think of any of these challenges 701 00:39:37,301 --> 00:39:39,011 and what could be the solution? 702 00:39:39,031 --> 00:39:44,441 It doesn't have to be really, detailed tech solution, but whatever you can think 703 00:39:44,441 --> 00:39:48,681 of, reach out, tell us what you think, what the solution could be, and we could 704 00:39:48,681 --> 00:39:50,956 be bounce ideas off of each other. 705 00:39:50,996 --> 00:39:52,986 Tell us what your idea is. 706 00:39:52,986 --> 00:39:55,576 And we could say, okay, this seems plausible. 707 00:39:55,576 --> 00:39:57,406 You should, pursue it more. 708 00:39:57,406 --> 00:40:00,206 So it would be an exciting thing to do. 709 00:40:00,206 --> 00:40:05,416 we look forward to getting your feedback and communication on how we would 710 00:40:05,416 --> 00:40:11,396 solve a lot of these challenges that we talked about today Thank you very much. 711 00:40:11,396 --> 00:40:13,966 Very much enjoyed talking about data centers. 712 00:40:13,976 --> 00:40:15,536 Hope our audience did the same. 713 00:40:15,996 --> 00:40:19,336 We look forward to getting your feedback and to hear from you. 714 00:40:19,446 --> 00:40:21,616 with that thought I'll end here. 715 00:40:21,736 --> 00:40:23,256 Thank you everyone for listening. 716 00:40:23,486 --> 00:40:24,226 Amit: Thanks everyone. 717 00:40:24,806 --> 00:40:25,196 Bye. 718 00:40:26,256 --> 00:40:26,706 Rinat: Bye.