1 1 00:00:00,310 --> 00:00:02,180 Data center architecture. 2 2 00:00:02,180 --> 00:00:04,500 In this lesson, we're going to discuss the architecture 3 3 00:00:04,500 --> 00:00:05,980 within our data centers. 4 4 00:00:05,980 --> 00:00:07,450 When we're talking about a data center, 5 5 00:00:07,450 --> 00:00:09,380 this is any facility that's composed of network, 6 6 00:00:09,380 --> 00:00:11,240 computers and storage that businesses 7 7 00:00:11,240 --> 00:00:12,870 and other organizations are going to use 8 8 00:00:12,870 --> 00:00:14,860 to organize, process, store, 9 9 00:00:14,860 --> 00:00:16,940 and disseminate large amounts of data. 10 10 00:00:16,940 --> 00:00:19,000 Now that's a pretty generic definition, 11 11 00:00:19,000 --> 00:00:19,963 but this is because a data center 12 12 00:00:19,963 --> 00:00:22,780 is used to describe a lot of different things these days 13 13 00:00:22,780 --> 00:00:25,490 from the very small to the massively large. 14 14 00:00:25,490 --> 00:00:27,340 For example, one of the smaller organizations 15 15 00:00:27,340 --> 00:00:29,160 I worked for had a small data center 16 16 00:00:29,160 --> 00:00:31,690 that was roughly 150 square feet in size. 17 17 00:00:31,690 --> 00:00:34,030 Within it, we had a single rack of networking equipment 18 18 00:00:34,030 --> 00:00:36,650 and about five racks that contained various servers. 19 19 00:00:36,650 --> 00:00:37,540 Now, on the other hand, 20 20 00:00:37,540 --> 00:00:39,220 one of the largest data centers in the world 21 21 00:00:39,220 --> 00:00:41,270 is located in Bluffdale, Utah. 22 22 00:00:41,270 --> 00:00:43,640 This data center is known as the Utah Data Center 23 23 00:00:43,640 --> 00:00:45,140 or its official title, 24 24 00:00:45,140 --> 00:00:47,200 the Intelligence Community Comprehensive National 25 25 00:00:47,200 --> 00:00:48,850 Cybersecurity Initiative Data Center, 26 26 00:00:48,850 --> 00:00:51,770 and it is 1.5 million square feet in size 27 27 00:00:51,770 --> 00:00:53,580 spread across 20 buildings. 28 28 00:00:53,580 --> 00:00:55,740 This data center is massive 29 29 00:00:55,740 --> 00:00:58,740 and it uses around 65 megawatts of electricity 30 30 00:00:58,740 --> 00:01:00,350 just to run it each day. 31 31 00:01:00,350 --> 00:01:02,120 Now this data center costs the US government 32 32 00:01:02,120 --> 00:01:04,730 about $1.5 billion to build 33 33 00:01:04,730 --> 00:01:06,930 and they spend over $40 million each year 34 34 00:01:06,930 --> 00:01:08,670 just on the electricity bills. 35 35 00:01:08,670 --> 00:01:10,730 This is a massive data center. 36 36 00:01:10,730 --> 00:01:13,000 Now, personally, I haven't worked at that data center, 37 37 00:01:13,000 --> 00:01:15,140 but I've worked at some really large data centers. 38 38 00:01:15,140 --> 00:01:17,240 But nothing at that size or scale. 39 39 00:01:17,240 --> 00:01:18,540 Now, Amazon, for example, 40 40 00:01:18,540 --> 00:01:20,300 uses data centers that range in size 41 41 00:01:20,300 --> 00:01:23,740 between 150,000 to 215,000 square feet. 42 42 00:01:23,740 --> 00:01:26,490 And they host about 50,000 to 80,000 servers 43 43 00:01:26,490 --> 00:01:28,160 in each of their data centers. 44 44 00:01:28,160 --> 00:01:29,400 As a network technician, 45 45 00:01:29,400 --> 00:01:30,720 you may very well end up working 46 46 00:01:30,720 --> 00:01:32,340 at one of these data centers one day. 47 47 00:01:32,340 --> 00:01:35,510 So let's explore their architecture just a little bit. 48 48 00:01:35,510 --> 00:01:38,340 In this lesson, we're going to talk about five key areas. 49 49 00:01:38,340 --> 00:01:40,630 The three-tiered hierarchy used by data centers, 50 50 00:01:40,630 --> 00:01:42,710 software-defined networking that allows our data centers 51 51 00:01:42,710 --> 00:01:43,990 to operate effectively, 52 52 00:01:43,990 --> 00:01:46,600 the spine and leaf architecture that's used by data centers, 53 53 00:01:46,600 --> 00:01:48,490 the traffic flows that are used by our data centers, 54 54 00:01:48,490 --> 00:01:50,800 and the on-premise versus hosted data centers 55 55 00:01:50,800 --> 00:01:52,650 and what the differences between them. 56 56 00:01:52,650 --> 00:01:55,360 So first let's look at the three-tiered hierarchy 57 57 00:01:55,360 --> 00:01:56,680 which consists of the core, 58 58 00:01:56,680 --> 00:01:58,490 the distribution or aggregation layer, 59 59 00:01:58,490 --> 00:02:00,520 and then the access or edge layer. 60 60 00:02:00,520 --> 00:02:02,700 The core layer is going to consist of the biggest 61 61 00:02:02,700 --> 00:02:04,670 and fastest and most expensive routers 62 62 00:02:04,670 --> 00:02:06,400 that you're ever going to end up working with. 63 63 00:02:06,400 --> 00:02:08,410 The core layer is going to be considered the backbone 64 64 00:02:08,410 --> 00:02:09,243 of our network 65 65 00:02:09,243 --> 00:02:11,550 and it's used to merge geographically separated networks 66 66 00:02:11,550 --> 00:02:13,800 into one logical and cohesive unit. 67 67 00:02:13,800 --> 00:02:15,910 In general, you're going to have at least two routers 68 68 00:02:15,910 --> 00:02:18,990 at the core level operating in a redundant configuration. 69 69 00:02:18,990 --> 00:02:20,900 After all, if you only had one core router 70 70 00:02:20,900 --> 00:02:21,760 and it went offline, 71 71 00:02:21,760 --> 00:02:24,420 the entire network would grind to a screeching halt. 72 72 00:02:24,420 --> 00:02:26,070 So we don't want to do that. 73 73 00:02:26,070 --> 00:02:29,100 Next, we have the distribution or aggregation layer. 74 74 00:02:29,100 --> 00:02:31,090 This layer is located under the core layer 75 75 00:02:31,090 --> 00:02:32,900 and it's going to provide boundary definition 76 76 00:02:32,900 --> 00:02:35,520 by implementing access control lists and filters. 77 77 00:02:35,520 --> 00:02:38,170 Now here at the distribution or aggregation layer, 78 78 00:02:38,170 --> 00:02:39,520 we're going to be defining policies 79 79 00:02:39,520 --> 00:02:41,060 for the network at large. 80 80 00:02:41,060 --> 00:02:43,010 Normally, you're going to see layer three switches 81 81 00:02:43,010 --> 00:02:45,260 here being used because this distribution layer 82 82 00:02:45,260 --> 00:02:47,310 is going to ensure packets are being properly routed 83 83 00:02:47,310 --> 00:02:49,130 between different subnets and VLANs 84 84 00:02:49,130 --> 00:02:51,020 within your enterprise network. 85 85 00:02:51,020 --> 00:02:54,030 Finally, we're going to get down to the access or edge layer. 86 86 00:02:54,030 --> 00:02:56,050 This layer is located beneath the distribution 87 87 00:02:56,050 --> 00:02:57,090 or aggregation layer, 88 88 00:02:57,090 --> 00:02:58,430 and it's going to be used to connect 89 89 00:02:58,430 --> 00:03:00,860 to all of your endpoint devices like your computers, 90 90 00:03:00,860 --> 00:03:03,100 your laptops, your servers, your printers, 91 91 00:03:03,100 --> 00:03:05,660 your wireless access points, and everything else. 92 92 00:03:05,660 --> 00:03:07,420 These access or edge devices 93 93 00:03:07,420 --> 00:03:09,170 are going to usually be regular switches, 94 94 00:03:09,170 --> 00:03:10,590 and they're going to be used to ensure packets 95 95 00:03:10,590 --> 00:03:11,990 are being converted to frames 96 96 00:03:11,990 --> 00:03:15,070 and delivered to the correct end point devices when needed. 97 97 00:03:15,070 --> 00:03:16,170 Now you may be wondering 98 98 00:03:16,170 --> 00:03:18,760 why do I need to use this three-tiered hierarchy? 99 99 00:03:18,760 --> 00:03:20,560 Well, by using this type of hierarchy, 100 100 00:03:20,560 --> 00:03:23,260 we can get better performance, management, scalability, 101 101 00:03:23,260 --> 00:03:25,000 and redundancy from our networks. 102 102 00:03:25,000 --> 00:03:26,330 It also is going to give us a better way 103 103 00:03:26,330 --> 00:03:27,540 to troubleshoot our network 104 104 00:03:27,540 --> 00:03:29,570 because normally if we find an issue, 105 105 00:03:29,570 --> 00:03:31,980 we're going to find it and isolate it down to a single access 106 106 00:03:31,980 --> 00:03:33,490 or edge layer device. 107 107 00:03:33,490 --> 00:03:35,430 Now, we can work on fixing that 108 108 00:03:35,430 --> 00:03:37,330 while the rest of the network keeps continuing to operate 109 109 00:03:37,330 --> 00:03:38,600 with no issues at all 110 110 00:03:38,600 --> 00:03:40,750 and then we can get our network all the way back up 111 111 00:03:40,750 --> 00:03:42,500 and running 100%. 112 112 00:03:42,500 --> 00:03:44,390 Normally inside your data center, 113 113 00:03:44,390 --> 00:03:46,200 you're going to find your core layer devices 114 114 00:03:46,200 --> 00:03:48,820 as well as your distribution or aggregation layer devices 115 115 00:03:48,820 --> 00:03:50,920 for the local network in that building. 116 116 00:03:50,920 --> 00:03:53,400 If you have remote branch offices or other locations, 117 117 00:03:53,400 --> 00:03:55,350 each one of those is going to have its own distribution 118 118 00:03:55,350 --> 00:03:57,380 or aggregation layer switch inside 119 119 00:03:57,380 --> 00:03:59,180 its main distribution frame too. 120 120 00:03:59,180 --> 00:04:00,150 In both cases, 121 121 00:04:00,150 --> 00:04:01,440 your network will then branch out 122 122 00:04:01,440 --> 00:04:03,140 to the intermediate distribution frames 123 123 00:04:03,140 --> 00:04:05,930 where you're going to find your access or edge layer devices. 124 124 00:04:05,930 --> 00:04:08,160 Now, when we talk about this three-tiered model, 125 125 00:04:08,160 --> 00:04:09,660 this is a traditional network 126 126 00:04:09,660 --> 00:04:11,830 you're going to find in most enterprises. 127 127 00:04:11,830 --> 00:04:15,360 Next, let's move into software-defined networking or SDN. 128 128 00:04:15,360 --> 00:04:16,480 Software-defined networking 129 129 00:04:16,480 --> 00:04:18,960 is a network architecture approach that enables the network 130 130 00:04:18,960 --> 00:04:21,130 to intelligently and centrally be controlled 131 131 00:04:21,130 --> 00:04:23,740 or programmed using software applications. 132 132 00:04:23,740 --> 00:04:25,020 This helps operators manage 133 133 00:04:25,020 --> 00:04:27,650 the entire network consistently and holistically, 134 134 00:04:27,650 --> 00:04:30,210 regardless of the underlying network technology. 135 135 00:04:30,210 --> 00:04:32,420 Essentially, we're going to take our physical networks 136 136 00:04:32,420 --> 00:04:34,190 and we can completely virtualize them 137 137 00:04:34,190 --> 00:04:35,790 or create a layer of abstraction 138 138 00:04:35,790 --> 00:04:38,280 between the physical devices and the logical architecture 139 139 00:04:38,280 --> 00:04:39,750 that they're going to represent. 140 140 00:04:39,750 --> 00:04:41,390 Now with software-defined networking, 141 141 00:04:41,390 --> 00:04:44,350 we can create complex networks very quickly and easily, 142 142 00:04:44,350 --> 00:04:47,480 leveraging increased network size and expanding their scope 143 143 00:04:47,480 --> 00:04:49,750 as well as their ability to rapidly change. 144 144 00:04:49,750 --> 00:04:51,190 In fact, one of the great things 145 145 00:04:51,190 --> 00:04:52,570 about software-defined networks 146 146 00:04:52,570 --> 00:04:54,050 is that they can be changed automatically 147 147 00:04:54,050 --> 00:04:57,360 by the network itself using automation and orchestration. 148 148 00:04:57,360 --> 00:04:58,310 On the other hand, 149 149 00:04:58,310 --> 00:05:00,580 all of this rapid change does make it harder for us 150 150 00:05:00,580 --> 00:05:02,190 as humans to keep up with it 151 151 00:05:02,190 --> 00:05:04,300 and fully understand the data flows on our network 152 152 00:05:04,300 --> 00:05:05,500 at any given time. 153 153 00:05:05,500 --> 00:05:07,460 Now, when it comes to software-defined networking, 154 154 00:05:07,460 --> 00:05:09,620 there are several pieces that we need to consider 155 155 00:05:09,620 --> 00:05:12,010 including the application layer, the control layer, 156 156 00:05:12,010 --> 00:05:14,840 the infrastructure layer, and the management plane. 157 157 00:05:14,840 --> 00:05:16,580 These three layers are going to allow the network 158 158 00:05:16,580 --> 00:05:19,270 to be decoupled from the underlying hardware itself. 159 159 00:05:19,270 --> 00:05:20,800 The application layer is going to focus 160 160 00:05:20,800 --> 00:05:22,540 on the communication resource requests 161 161 00:05:22,540 --> 00:05:25,120 or information about the network as a whole. 162 162 00:05:25,120 --> 00:05:27,230 The control layer is then going to use that information 163 163 00:05:27,230 --> 00:05:29,940 from the applications and decide how to route a data packet 164 164 00:05:29,940 --> 00:05:31,130 on that network. 165 165 00:05:31,130 --> 00:05:32,680 It also makes decisions about how traffic 166 166 00:05:32,680 --> 00:05:33,760 should be prioritized, 167 167 00:05:33,760 --> 00:05:34,830 how it should be secured, 168 168 00:05:34,830 --> 00:05:36,640 and where it should be forwarded to. 169 169 00:05:36,640 --> 00:05:38,250 The infrastructure layer is going to contain 170 170 00:05:38,250 --> 00:05:40,850 the actual networking devices that receive the information 171 171 00:05:40,850 --> 00:05:43,090 from the control layer about where to move the data, 172 172 00:05:43,090 --> 00:05:45,380 and then it's going to perform those movements. 173 173 00:05:45,380 --> 00:05:47,180 Now with software-defined networking, 174 174 00:05:47,180 --> 00:05:49,930 these underlying infrastructure devices can be physical 175 175 00:05:49,930 --> 00:05:52,620 or virtual devices depending on your network configuration 176 176 00:05:52,620 --> 00:05:55,060 or a combination or hybrid of the two. 177 177 00:05:55,060 --> 00:05:57,780 Remember the whole concept with software-defined networking, 178 178 00:05:57,780 --> 00:05:59,390 is probably this layer of abstraction 179 179 00:05:59,390 --> 00:06:01,060 between the real underlying devices 180 180 00:06:01,060 --> 00:06:02,370 and the control and data flow 181 181 00:06:02,370 --> 00:06:04,270 that's going to happen on that network. 182 182 00:06:04,270 --> 00:06:06,410 Software-defined networking is also critical for use 183 183 00:06:06,410 --> 00:06:08,180 in our successful cloud applications 184 184 00:06:08,180 --> 00:06:10,730 because it gives us the scalability and elasticity 185 185 00:06:10,730 --> 00:06:13,720 and agility that we need inside those networks. 186 186 00:06:13,720 --> 00:06:15,810 Now, the fourth part of software-defined networks 187 187 00:06:15,810 --> 00:06:17,260 is our management plane. 188 188 00:06:17,260 --> 00:06:18,700 Now the management plane is going to be used 189 189 00:06:18,700 --> 00:06:21,430 to monitor traffic conditions and the status of the network. 190 190 00:06:21,430 --> 00:06:23,360 Basically, the management plane is going to allow us 191 191 00:06:23,360 --> 00:06:26,470 to oversee the network and gain insight into its operations. 192 192 00:06:26,470 --> 00:06:28,720 This will also allow us to make configuration changes, 193 193 00:06:28,720 --> 00:06:30,320 to set things up the way we want, 194 194 00:06:30,320 --> 00:06:33,170 and make sure they're working the way we need them to. 195 195 00:06:33,170 --> 00:06:34,470 So for the exam, 196 196 00:06:34,470 --> 00:06:36,590 I want you to remember that software-defined networking 197 197 00:06:36,590 --> 00:06:39,680 or SDN is broken down into four parts. 198 198 00:06:39,680 --> 00:06:41,700 The application layer, the control layer, 199 199 00:06:41,700 --> 00:06:44,550 the infrastructure layer, and the management plane. 200 200 00:06:44,550 --> 00:06:47,570 Next, let's discuss the spine and leaf architecture. 201 201 00:06:47,570 --> 00:06:49,040 Now, the spine and leaf architecture 202 202 00:06:49,040 --> 00:06:51,070 is an alternative type of network architecture 203 203 00:06:51,070 --> 00:06:53,570 that's used specifically within the data center. 204 204 00:06:53,570 --> 00:06:55,660 Now, when we use our three-tiered hierarchy 205 205 00:06:55,660 --> 00:06:56,493 that we covered earlier, 206 206 00:06:56,493 --> 00:06:57,860 we talked about the fact that we connected 207 207 00:06:57,860 --> 00:07:00,020 the core layer down to the distribution layer 208 208 00:07:00,020 --> 00:07:02,570 and then down to the access layer or edge layer 209 209 00:07:02,570 --> 00:07:04,480 with all of our end point devices. 210 210 00:07:04,480 --> 00:07:06,120 With a spine and leaf architecture, 211 211 00:07:06,120 --> 00:07:08,070 instead we're going to be focused on communication 212 212 00:07:08,070 --> 00:07:10,700 within the data center itself only specifically 213 213 00:07:10,700 --> 00:07:12,920 to the server firm and the portions of it. 214 214 00:07:12,920 --> 00:07:14,640 The spine and leaf architecture consists 215 215 00:07:14,640 --> 00:07:17,810 of two switching layers known as a spine and a leaf. 216 216 00:07:17,810 --> 00:07:18,690 Now the leaf layer 217 217 00:07:18,690 --> 00:07:20,640 is going to consist of all the access switches 218 218 00:07:20,640 --> 00:07:22,770 that aggregate traffic from the different servers 219 219 00:07:22,770 --> 00:07:24,880 and then connect directly into the spine layer 220 220 00:07:24,880 --> 00:07:26,470 or the networks core. 221 221 00:07:26,470 --> 00:07:28,130 This spine contains the switches 222 222 00:07:28,130 --> 00:07:30,010 that will interconnect all the leaf layer switches 223 223 00:07:30,010 --> 00:07:32,010 into a full mesh topology. 224 224 00:07:32,010 --> 00:07:34,420 This leads to increased performance and redundancy 225 225 00:07:34,420 --> 00:07:36,610 for all the servers that are connected to the leaf layer 226 226 00:07:36,610 --> 00:07:38,650 and in turn to the spine layer. 227 227 00:07:38,650 --> 00:07:39,950 Now, as you may be wondering, 228 228 00:07:39,950 --> 00:07:41,910 why did I move from the three-tiered networks 229 229 00:07:41,910 --> 00:07:44,540 to software-defined networks before starting to discuss 230 230 00:07:44,540 --> 00:07:46,160 the spine and leaf architecture? 231 231 00:07:46,160 --> 00:07:48,140 Well, this is because many of our spine 232 232 00:07:48,140 --> 00:07:50,610 and leaf architectures rely on software-defined networks 233 233 00:07:50,610 --> 00:07:51,530 to operate. 234 234 00:07:51,530 --> 00:07:53,270 By using a spine and leaf architecture, 235 235 00:07:53,270 --> 00:07:55,690 we can actually get faster speeds and lower latency 236 236 00:07:55,690 --> 00:07:58,110 than the traditional three-tiered hierarchy as well. 237 237 00:07:58,110 --> 00:07:59,900 By using a spine and leaf architecture, 238 238 00:07:59,900 --> 00:08:01,030 we can actually take shortcuts 239 239 00:08:01,030 --> 00:08:02,520 in getting data from place to place, 240 240 00:08:02,520 --> 00:08:03,800 and this all happens best 241 241 00:08:03,800 --> 00:08:05,500 when we're using software-defined networks 242 242 00:08:05,500 --> 00:08:07,970 in combination with a spine and leaf design. 243 243 00:08:07,970 --> 00:08:10,160 If you're installing a spine and leaf architecture, 244 244 00:08:10,160 --> 00:08:11,673 normally you're going to install two switches 245 245 00:08:11,673 --> 00:08:13,280 into each server rack. 246 246 00:08:13,280 --> 00:08:15,210 This is known as top of rack switching 247 247 00:08:15,210 --> 00:08:16,730 because the switches are physically installed 248 248 00:08:16,730 --> 00:08:18,340 at the very top of the server rack, 249 249 00:08:18,340 --> 00:08:19,990 and each server inside that rack 250 250 00:08:19,990 --> 00:08:21,870 will have a connection to each of the two switches 251 251 00:08:21,870 --> 00:08:23,810 that are residing inside that rack. 252 252 00:08:23,810 --> 00:08:25,390 Now, these switches are essentially going to be 253 253 00:08:25,390 --> 00:08:27,870 the leafs inside our spine and leaf architecture. 254 254 00:08:27,870 --> 00:08:29,530 And they're going to connect back to the spine 255 255 00:08:29,530 --> 00:08:30,740 which is going to serve as the backbone 256 256 00:08:30,740 --> 00:08:32,440 of our data center network. 257 257 00:08:32,440 --> 00:08:34,700 This spine and leaf architecture can also be combined 258 258 00:08:34,700 --> 00:08:36,710 with our standard three-tier hierarchy. 259 259 00:08:36,710 --> 00:08:37,830 Now, when we do this, 260 260 00:08:37,830 --> 00:08:39,590 all the servers in the data center will connect 261 261 00:08:39,590 --> 00:08:40,600 to the leaf layers, 262 262 00:08:40,600 --> 00:08:42,750 and the leaf layers will connect to the spine. 263 263 00:08:42,750 --> 00:08:44,540 But the spine will then connect directly 264 264 00:08:44,540 --> 00:08:46,640 to the core layer of the three-tiered model 265 265 00:08:46,640 --> 00:08:49,130 whereas all the other non-data center devices 266 266 00:08:49,130 --> 00:08:50,630 will connect to the access layer 267 267 00:08:50,630 --> 00:08:52,830 and then up to the distribution or aggregation layers 268 268 00:08:52,830 --> 00:08:55,020 before connecting into the core layer. 269 269 00:08:55,020 --> 00:08:56,980 Next, we need to discuss traffic flows 270 270 00:08:56,980 --> 00:08:58,680 in relation to our data centers. 271 271 00:08:58,680 --> 00:08:59,990 Now there's two main types. 272 272 00:08:59,990 --> 00:09:02,310 We have North-South and East-West. 273 273 00:09:02,310 --> 00:09:03,620 These two terms are used to describe 274 274 00:09:03,620 --> 00:09:06,910 the direction of traffic flow into or out of a data center. 275 275 00:09:06,910 --> 00:09:08,590 Now, when we have North-South traffic, 276 276 00:09:08,590 --> 00:09:10,600 this is going to refer to communication traffic 277 277 00:09:10,600 --> 00:09:12,070 that enters or leaves the data center 278 278 00:09:12,070 --> 00:09:15,090 from a system fiscally residing outside of the data center. 279 279 00:09:15,090 --> 00:09:17,560 So when we talk specifically about North traffic, 280 280 00:09:17,560 --> 00:09:20,100 this is traffic that is exiting your data center. 281 281 00:09:20,100 --> 00:09:21,780 Southbound traffic on the other hand 282 282 00:09:21,780 --> 00:09:24,360 is referring to traffic that is entering your data center. 283 283 00:09:24,360 --> 00:09:26,620 In both cases, this data is exiting 284 284 00:09:26,620 --> 00:09:29,130 or entering the data center going through a firewall 285 285 00:09:29,130 --> 00:09:31,170 or other network infrastructure boundary device 286 286 00:09:31,170 --> 00:09:32,460 such as a router. 287 287 00:09:32,460 --> 00:09:34,980 Conversely, we also have East-West traffic. 288 288 00:09:34,980 --> 00:09:36,960 Now, East-West traffic refers to data flow 289 289 00:09:36,960 --> 00:09:38,380 within a data center. 290 290 00:09:38,380 --> 00:09:41,070 For example, if we're using a spine and leaf architecture, 291 291 00:09:41,070 --> 00:09:43,840 any data flow between various servers in the data center, 292 292 00:09:43,840 --> 00:09:45,550 even if it goes between different leafs 293 293 00:09:45,550 --> 00:09:47,670 would be considered East-West traffic 294 294 00:09:47,670 --> 00:09:50,490 because that data is not leaving our data center. 295 295 00:09:50,490 --> 00:09:53,020 Now due to the increased use of software-defined networking, 296 296 00:09:53,020 --> 00:09:55,810 virtualization, private cloud, and converge networks, 297 297 00:09:55,810 --> 00:09:57,420 more and more traffic that we're using 298 298 00:09:57,420 --> 00:09:59,870 is being classified as East-West traffic, 299 299 00:09:59,870 --> 00:10:02,430 because it's still virtually part of your data center. 300 300 00:10:02,430 --> 00:10:03,540 So in summary, 301 301 00:10:03,540 --> 00:10:05,530 if the data is entering the data center, 302 302 00:10:05,530 --> 00:10:07,400 it's considered Southbound traffic. 303 303 00:10:07,400 --> 00:10:08,790 If it's leaving the data center, 304 304 00:10:08,790 --> 00:10:10,470 it's considered Northbound traffic. 305 305 00:10:10,470 --> 00:10:12,000 If it's moving within the data center, 306 306 00:10:12,000 --> 00:10:13,950 it's considered East-West traffic. 307 307 00:10:13,950 --> 00:10:15,870 Finally, we need to talk about on-premise 308 308 00:10:15,870 --> 00:10:17,640 versus hosted data centers. 309 309 00:10:17,640 --> 00:10:19,140 This discussion is really going to come down 310 310 00:10:19,140 --> 00:10:20,940 to where are you going to store your data? 311 311 00:10:20,940 --> 00:10:23,090 Now, if you're using an on-premise data center, 312 312 00:10:23,090 --> 00:10:25,650 you're using a traditional private data infrastructure 313 313 00:10:25,650 --> 00:10:27,680 where your organization has its own data center 314 314 00:10:27,680 --> 00:10:29,900 that houses all of its servers and networking equipment 315 315 00:10:29,900 --> 00:10:32,490 that it's going to use to be able to support its operations. 316 316 00:10:32,490 --> 00:10:34,920 We call this on-premise because it's usually located 317 317 00:10:34,920 --> 00:10:37,090 in the same building as your main office. 318 318 00:10:37,090 --> 00:10:39,330 Sometimes though, you're going to have multiple offices 319 319 00:10:39,330 --> 00:10:42,940 spread across a large geographic region or across the globe. 320 320 00:10:42,940 --> 00:10:45,290 Normally, one of your offices is going to be considered 321 321 00:10:45,290 --> 00:10:47,820 your headquarters and will host your on-premise data center 322 322 00:10:47,820 --> 00:10:49,310 in this type of organization. 323 323 00:10:49,310 --> 00:10:51,400 And then all the other offices around the globe, 324 324 00:10:51,400 --> 00:10:53,240 they'll be called branch offices. 325 325 00:10:53,240 --> 00:10:54,470 Now these branch offices 326 326 00:10:54,470 --> 00:10:56,200 usually will not host their own servers, 327 327 00:10:56,200 --> 00:10:57,560 but instead they will host them 328 328 00:10:57,560 --> 00:11:00,280 in your on-premise data center at your headquarters. 329 329 00:11:00,280 --> 00:11:01,360 If you have a fast enough connection 330 330 00:11:01,360 --> 00:11:03,330 between each branch office and the headquarters, 331 331 00:11:03,330 --> 00:11:05,660 you can host everything at your main data center. 332 332 00:11:05,660 --> 00:11:07,460 But if you have slower connections, 333 333 00:11:07,460 --> 00:11:09,310 you may need to host some services locally 334 334 00:11:09,310 --> 00:11:11,130 inside the branch office too. 335 335 00:11:11,130 --> 00:11:13,020 For example, when I was an IT director 336 336 00:11:13,020 --> 00:11:15,330 for an organization spread across multiple countries, 337 337 00:11:15,330 --> 00:11:18,160 we had our own on-premise data center at our headquarters 338 338 00:11:18,160 --> 00:11:19,780 where we hosted most of the services 339 339 00:11:19,780 --> 00:11:22,530 including our domain controllers, email, proxy servers, 340 340 00:11:22,530 --> 00:11:23,780 and other services, 341 341 00:11:23,780 --> 00:11:25,910 but we still maintain a local file server 342 342 00:11:25,910 --> 00:11:28,740 in each branch office for them to use for their share drives 343 343 00:11:28,740 --> 00:11:31,030 because otherwise they would be transferring gigabytes 344 344 00:11:31,030 --> 00:11:33,600 and gigabytes of data to and from our data centers 345 345 00:11:33,600 --> 00:11:35,000 over a small connection. 346 346 00:11:35,000 --> 00:11:37,510 And this would really slow down the entire network. 347 347 00:11:37,510 --> 00:11:39,570 Now, this solution for us worked really well 348 348 00:11:39,570 --> 00:11:40,980 because the branch office locations 349 349 00:11:40,980 --> 00:11:43,760 only need to access their own files on their share drive 350 350 00:11:43,760 --> 00:11:45,760 not any files from the main office. 351 351 00:11:45,760 --> 00:11:47,830 Now, if they instead need to have a shared file server 352 352 00:11:47,830 --> 00:11:49,120 with us at the head office, 353 353 00:11:49,120 --> 00:11:50,710 we would have opted for something else 354 354 00:11:50,710 --> 00:11:52,820 like a content engine or a caching engine 355 355 00:11:52,820 --> 00:11:54,400 instead of using a local file server 356 356 00:11:54,400 --> 00:11:55,780 to meet their needs. 357 357 00:11:55,780 --> 00:11:56,990 Now, the other option though, 358 358 00:11:56,990 --> 00:11:59,830 is to host your data using a co-located data center. 359 359 00:11:59,830 --> 00:12:02,030 Now in a co-located data center arrangement, 360 360 00:12:02,030 --> 00:12:03,710 your organization places their servers 361 361 00:12:03,710 --> 00:12:05,910 and networking equipment in the data center environment 362 362 00:12:05,910 --> 00:12:07,550 that's owned by another company. 363 363 00:12:07,550 --> 00:12:10,140 Essentially, you're going to rent space in their data center 364 364 00:12:10,140 --> 00:12:11,840 instead of having to build your own. 365 365 00:12:11,840 --> 00:12:14,690 When I started my very first company back in 1999, 366 366 00:12:14,690 --> 00:12:16,830 I was doing website design, website hosting, 367 367 00:12:16,830 --> 00:12:18,270 and building network architectures 368 368 00:12:18,270 --> 00:12:20,300 for small and medium sized businesses. 369 369 00:12:20,300 --> 00:12:21,460 Because I was just starting out, 370 370 00:12:21,460 --> 00:12:23,660 I couldn't afford the millions of dollars of the cost 371 371 00:12:23,660 --> 00:12:25,150 to make my own data center. 372 372 00:12:25,150 --> 00:12:27,110 So we wanted to host our web servers 373 373 00:12:27,110 --> 00:12:28,820 and we decided to lease two racks 374 374 00:12:28,820 --> 00:12:30,620 in a larger company's data center. 375 375 00:12:30,620 --> 00:12:32,080 For a fixed price each month, 376 376 00:12:32,080 --> 00:12:33,740 I was able to put whatever equipment I wanted 377 377 00:12:33,740 --> 00:12:35,350 that would fit into those two racks 378 378 00:12:35,350 --> 00:12:37,400 and they would provide me with a Cat5 connection 379 379 00:12:37,400 --> 00:12:39,320 with a hundred megabits per second of bandwidth, 380 380 00:12:39,320 --> 00:12:41,570 power, and backup batteries, and generators. 381 381 00:12:41,570 --> 00:12:44,200 Beyond that it was up to me to drive their facility, 382 382 00:12:44,200 --> 00:12:45,720 install and maintain all my own servers 383 383 00:12:45,720 --> 00:12:46,650 and networking equipment 384 384 00:12:46,650 --> 00:12:48,930 and run all the configurations I needed. 385 385 00:12:48,930 --> 00:12:50,400 Now, the last option we have 386 386 00:12:50,400 --> 00:12:52,700 is to move everything into a cloud-based platform 387 387 00:12:52,700 --> 00:12:55,540 like Amazon Web Services or Microsoft Azure. 388 388 00:12:55,540 --> 00:12:57,800 In those cases we can't use co-location 389 389 00:12:57,800 --> 00:12:59,490 because they're not going to allow us put our own servers 390 390 00:12:59,490 --> 00:13:00,800 into their facilities. 391 391 00:13:00,800 --> 00:13:02,750 Instead we would have to migrate all of our data 392 392 00:13:02,750 --> 00:13:04,650 out of our servers and our data centers 393 393 00:13:04,650 --> 00:13:07,290 and put them into their servers and their data centers. 394 394 00:13:07,290 --> 00:13:09,630 All right, as you can see there are lots of different ways 395 395 00:13:09,630 --> 00:13:12,000 to architect your data centers and your networks. 396 396 00:13:12,000 --> 00:13:14,090 Everything from a three-tiered traditional model 397 397 00:13:14,090 --> 00:13:15,860 to a more modern software-defined network 398 398 00:13:15,860 --> 00:13:18,050 that implements spine and leafs to moving completely 399 399 00:13:18,050 --> 00:13:18,910 to the cloud. 400 400 00:13:18,910 --> 00:13:20,560 It really depends on your business case 401 401 00:13:20,560 --> 00:13:22,060 and your organization's needs.