1 00:00:00,360 --> 00:00:04,230 ‫So now let's talk about the AWS Snow family. 2 00:00:04,230 --> 00:00:08,250 ‫So it represents a highly secure, portable devices 3 00:00:08,250 --> 00:00:11,760 ‫that has two use cases within AWS. 4 00:00:11,760 --> 00:00:15,120 ‫Either it's used to collect and process data at the edge 5 00:00:15,120 --> 00:00:19,650 ‫or to migrate data in and out of AWS. 6 00:00:19,650 --> 00:00:21,270 ‫So we have two use cases, the data migration, 7 00:00:21,270 --> 00:00:22,920 ‫and for this we have three different types 8 00:00:22,920 --> 00:00:24,840 ‫of devices within the Snow family. 9 00:00:24,840 --> 00:00:29,400 ‫We have the Snowcone, the Snowball Edge, and the Snowmobile. 10 00:00:29,400 --> 00:00:31,740 ‫And for the second use case, edge computing, 11 00:00:31,740 --> 00:00:34,680 ‫we have the Snowcone, and the Snowball Edge. 12 00:00:34,680 --> 00:00:37,140 ‫So we'll first tackle the data migration subject 13 00:00:37,140 --> 00:00:39,210 ‫and then the edge computing. 14 00:00:39,210 --> 00:00:41,880 ‫So why do we want to do data migration 15 00:00:41,880 --> 00:00:44,550 ‫with the AWS Snow family? 16 00:00:44,550 --> 00:00:46,860 ‫Well, if we look at the time it takes 17 00:00:46,860 --> 00:00:50,280 ‫to transfer a lot of data over the network, 18 00:00:50,280 --> 00:00:52,020 ‫it can take a lot of time. 19 00:00:52,020 --> 00:00:56,310 ‫So for example, if we want to transfer 100 terabytes 20 00:00:56,310 --> 00:01:00,120 ‫over a one gigabits per second network line, 21 00:01:00,120 --> 00:01:03,240 ‫it will take us 12 days to achieve it. 22 00:01:03,240 --> 00:01:04,073 ‫Okay? 23 00:01:04,073 --> 00:01:06,060 ‫And so obviously if we do a petabyte, 24 00:01:06,060 --> 00:01:07,740 ‫it will take forever, and so on. 25 00:01:07,740 --> 00:01:08,670 ‫So as we can see 26 00:01:08,670 --> 00:01:12,210 ‫sometimes we just want the data to get to AWS fast. 27 00:01:12,210 --> 00:01:14,040 ‫And the challenge is that sometimes 28 00:01:14,040 --> 00:01:17,310 ‫on top of having a small network transfer, 29 00:01:17,310 --> 00:01:18,780 ‫you have limited connectivity, 30 00:01:18,780 --> 00:01:21,330 ‫limited bitwidth, bandwidth, sorry, 31 00:01:21,330 --> 00:01:23,100 ‫transferring data over the network 32 00:01:23,100 --> 00:01:24,300 ‫may cost you some money, okay? 33 00:01:24,300 --> 00:01:26,790 ‫It's not free to use a network. 34 00:01:26,790 --> 00:01:28,770 ‫It could be that also the bandwidth is shared. 35 00:01:28,770 --> 00:01:32,250 ‫For example, if you download a video from AWS 36 00:01:32,250 --> 00:01:35,910 ‫and you download 10 terabytes of data, you know, 37 00:01:35,910 --> 00:01:38,400 ‫maybe you're going to block your entire office 38 00:01:38,400 --> 00:01:42,060 ‫because you're maximizing the bandwidth within your office, 39 00:01:42,060 --> 00:01:44,430 ‫and then maybe the connection is not stable enough 40 00:01:44,430 --> 00:01:46,710 ‫so you have to retry, and so on. 41 00:01:46,710 --> 00:01:51,000 ‫So all these reasons make a case for Snow family. 42 00:01:51,000 --> 00:01:53,460 ‫So the Snow family are offline devices 43 00:01:53,460 --> 00:01:55,770 ‫that allow you to perform data migrations. 44 00:01:55,770 --> 00:01:58,980 ‫So AWS will send you an actual physical device 45 00:01:58,980 --> 00:02:01,560 ‫by, you know, the post office, 46 00:02:01,560 --> 00:02:03,390 ‫and then you load your data onto it 47 00:02:03,390 --> 00:02:05,820 ‫and then you send it back to AWS. 48 00:02:05,820 --> 00:02:08,550 ‫So the rule of thumb is that if it takes more than a week 49 00:02:08,550 --> 00:02:10,440 ‫to transfer data over the network, 50 00:02:10,440 --> 00:02:13,920 ‫then you should use a Snowball device, for example. 51 00:02:13,920 --> 00:02:16,410 ‫So to really explain how that works, let's take an example. 52 00:02:16,410 --> 00:02:19,530 ‫If you wanted to directly upload a file into Amazon S3, 53 00:02:19,530 --> 00:02:22,110 ‫we have the clients sends the data into Amazon S3, 54 00:02:22,110 --> 00:02:23,460 ‫very easy, right? 55 00:02:23,460 --> 00:02:25,020 ‫But with the Snow family, 56 00:02:25,020 --> 00:02:26,790 ‫for example, with a Snowball device, 57 00:02:26,790 --> 00:02:28,980 ‫the client request a Snowball device, 58 00:02:28,980 --> 00:02:30,870 ‫we receive it via the post, okay? 59 00:02:30,870 --> 00:02:33,060 ‫AWS will deliver the device to us. 60 00:02:33,060 --> 00:02:36,630 ‫We load the data directly onto the devices locally, 61 00:02:36,630 --> 00:02:40,440 ‫and then we ship back the device to AWS, 62 00:02:40,440 --> 00:02:42,570 ‫into an AWS facility. 63 00:02:42,570 --> 00:02:44,460 ‫Then they will take the device 64 00:02:44,460 --> 00:02:47,460 ‫and they will plug it into their own infrastructure, 65 00:02:47,460 --> 00:02:49,980 ‫and then the data will be imported or exported, 66 00:02:49,980 --> 00:02:51,480 ‫based on what you want to do, 67 00:02:51,480 --> 00:02:54,570 ‫to an Amazon S3 bucket, and you're good to go. 68 00:02:54,570 --> 00:02:57,390 ‫So really it is a way to transfer data to AWS, 69 00:02:57,390 --> 00:03:01,318 ‫but through the physical route, not the network route. 70 00:03:01,318 --> 00:03:04,320 ‫Okay, so now what sort of devices do we have? 71 00:03:04,320 --> 00:03:06,150 ‫We have the Snowball Edge. 72 00:03:06,150 --> 00:03:09,150 ‫And Snowball Edge is a huge box as you can see, 73 00:03:09,150 --> 00:03:10,680 ‫and it is going to be used 74 00:03:10,680 --> 00:03:15,360 ‫to move terabytes or petabytes of data in and out of AWS. 75 00:03:15,360 --> 00:03:16,500 ‫It's going to be an alternative 76 00:03:16,500 --> 00:03:18,450 ‫to moving data over the network as we've seen. 77 00:03:18,450 --> 00:03:21,030 ‫We're going to pay per data transfer job. 78 00:03:21,030 --> 00:03:24,180 ‫And the interface within the Snowball Edge 79 00:03:24,180 --> 00:03:25,920 ‫is going to provide a block storage 80 00:03:25,920 --> 00:03:29,100 ‫or Amazon S3 compatible object storage. 81 00:03:29,100 --> 00:03:32,010 ‫So we have two flavors for the Snowball Edge. 82 00:03:32,010 --> 00:03:34,590 ‫We have the Snowball Edge Storage Optimized 83 00:03:34,590 --> 00:03:36,790 ‫which is going to give us 80 terabytes 84 00:03:38,520 --> 00:03:41,520 ‫of a hardware disk capacity, 85 00:03:41,520 --> 00:03:42,750 ‫which works for block volume 86 00:03:42,750 --> 00:03:44,640 ‫or S3 compatible object storage. 87 00:03:44,640 --> 00:03:47,700 ‫And we have the Snowball Edge Compute Optimized 88 00:03:47,700 --> 00:03:51,870 ‫of space of 42 TB or 28 TB. 89 00:03:51,870 --> 00:03:53,880 ‫So really, if we want to have more storage, 90 00:03:53,880 --> 00:03:55,290 ‫obviously we want to get the flavor 91 00:03:55,290 --> 00:03:59,010 ‫that is called Snowball Edge Storage Optimized. 92 00:03:59,010 --> 00:04:02,790 ‫So the use case for a Snowball Edge for data transfer so far 93 00:04:02,790 --> 00:04:05,370 ‫is to do a large data cloud migration 94 00:04:05,370 --> 00:04:07,140 ‫to decommission a data center 95 00:04:07,140 --> 00:04:08,820 ‫or maybe to do disaster recovery 96 00:04:08,820 --> 00:04:11,700 ‫by backing up your data into AWS. 97 00:04:11,700 --> 00:04:13,350 ‫Next we have the Snowcone. 98 00:04:13,350 --> 00:04:16,320 ‫So the Snowcone is a very small portable device. 99 00:04:16,320 --> 00:04:18,150 ‫And it's rugged, it's secure, 100 00:04:18,150 --> 00:04:20,310 ‫it can withstand harsh environment, 101 00:04:20,310 --> 00:04:22,500 ‫and it's meant for environments 102 00:04:22,500 --> 00:04:24,450 ‫where you have a little amount of data. 103 00:04:24,450 --> 00:04:26,400 ‫So it's light, it's like 2.1 kilograms, 104 00:04:26,400 --> 00:04:28,470 ‫and you can put it on a drone if you wanted to. 105 00:04:28,470 --> 00:04:29,520 ‫And it's going to be used 106 00:04:29,520 --> 00:04:32,730 ‫for edge computing, storage, and data transfer. 107 00:04:32,730 --> 00:04:33,840 ‫You have two flavors of it. 108 00:04:33,840 --> 00:04:34,673 ‫You have the Snowcone 109 00:04:34,673 --> 00:04:37,440 ‫which comes with eight terabytes of HDD storage, 110 00:04:37,440 --> 00:04:41,160 ‫and the Snowcone SSD of 14 terabytes of SSD storage 111 00:04:41,160 --> 00:04:42,720 ‫if you need a faster disk. 112 00:04:42,720 --> 00:04:45,540 ‫You will use the Snowcone where the Snowball does not fit, 113 00:04:45,540 --> 00:04:47,940 ‫for example, in the space constraint environment. 114 00:04:47,940 --> 00:04:50,760 ‫And you must provide your own batteries and cables. 115 00:04:50,760 --> 00:04:54,120 ‫Now, to send back the data to AWS, you have two options. 116 00:04:54,120 --> 00:04:57,300 ‫So either you send the data back offline by shipping it, 117 00:04:57,300 --> 00:04:59,130 ‫or you can connect this device, 118 00:04:59,130 --> 00:05:02,370 ‫after it has captured some data, into a data center, 119 00:05:02,370 --> 00:05:05,610 ‫for example, whatever that has an internet connection, 120 00:05:05,610 --> 00:05:08,580 ‫and then use the AWS DataSync service 121 00:05:08,580 --> 00:05:11,193 ‫to send the data back to AWS. 122 00:05:12,690 --> 00:05:15,600 ‫And then Snowmobile is an actual truck. 123 00:05:15,600 --> 00:05:16,980 ‫So when they announced it, 124 00:05:16,980 --> 00:05:18,660 ‫they actually took a truck on stage 125 00:05:18,660 --> 00:05:20,940 ‫to show that it was an actual truck, 126 00:05:20,940 --> 00:05:22,890 ‫that is going to transfer your data. 127 00:05:22,890 --> 00:05:24,300 ‫And so with the Snowmobile, 128 00:05:24,300 --> 00:05:26,190 ‫you can transfer exabytes of data. 129 00:05:26,190 --> 00:05:30,570 ‫So 1 exabyte is 1,000 petabytes, is 1 million terabytes. 130 00:05:30,570 --> 00:05:34,320 ‫And each Snowmobile will have 100 petabytes of capacity. 131 00:05:34,320 --> 00:05:36,900 ‫So if you wanted to reach one exabyte of data, 132 00:05:36,900 --> 00:05:39,930 ‫you need to order 10 Snowmobiles. 133 00:05:39,930 --> 00:05:42,000 ‫It's high security, it's temperature controlled, 134 00:05:42,000 --> 00:05:44,130 ‫it has GPS, 24/7 video surveillance. 135 00:05:44,130 --> 00:05:46,350 ‫So it's quite a secure way to transfer your data. 136 00:05:46,350 --> 00:05:49,140 ‫And it's a better use case than Snowball 137 00:05:49,140 --> 00:05:52,950 ‫if you start transferring more than 10 petabytes of data. 138 00:05:52,950 --> 00:05:56,820 ‫So as a summary, for data migration, we have three options. 139 00:05:56,820 --> 00:06:00,060 ‫We have Snowcone, Snowball Edge, Snowmobile. 140 00:06:00,060 --> 00:06:02,190 ‫And each come with different storage capacities. 141 00:06:02,190 --> 00:06:05,880 ‫So 8 terabytes, 80 terabytes, and 100 petabytes. 142 00:06:05,880 --> 00:06:08,610 ‫The migration size that's recommended by AWS is, 143 00:06:08,610 --> 00:06:11,820 ‫a Snowcone is up to 24 terabytes. 144 00:06:11,820 --> 00:06:14,370 ‫For Snowball Edge, it's up to petabytes, 145 00:06:14,370 --> 00:06:16,890 ‫and it's offline because you have to send it back to AWS. 146 00:06:16,890 --> 00:06:21,360 ‫And for Snowmobile, the use case is up to exabytes of data. 147 00:06:21,360 --> 00:06:23,670 ‫DataSync agent is pre-installed on a Snowcone 148 00:06:23,670 --> 00:06:25,350 ‫because you can plug it to your network 149 00:06:25,350 --> 00:06:26,790 ‫and have DataSync send the data 150 00:06:26,790 --> 00:06:28,860 ‫over the network to AWS as well. 151 00:06:28,860 --> 00:06:31,440 ‫And for Snowball Edge, you can do storage clustering. 152 00:06:31,440 --> 00:06:35,220 ‫So you can put up to 15 Snowball Edges together 153 00:06:35,220 --> 00:06:37,680 ‫to increase the storage size. 154 00:06:37,680 --> 00:06:41,610 ‫Okay, so how do we use a Snow family device? 155 00:06:41,610 --> 00:06:44,310 ‫Well, you request the device from the console for delivery 156 00:06:44,310 --> 00:06:45,930 ‫and we'll see this in the hands-on. 157 00:06:45,930 --> 00:06:47,580 ‫Then we install the Snowball clients 158 00:06:47,580 --> 00:06:49,710 ‫or we use AWS OpsHub, 159 00:06:49,710 --> 00:06:52,260 ‫that will see in this lecture, on your servers. 160 00:06:52,260 --> 00:06:54,150 ‫Then we connect the Snowball to the servers 161 00:06:54,150 --> 00:06:56,550 ‫and you start copying the files in your clients. 162 00:06:56,550 --> 00:06:58,440 ‫Then we ship back the device when we're ready. 163 00:06:58,440 --> 00:07:01,080 ‫It will go straight to the right AWS facility, 164 00:07:01,080 --> 00:07:03,180 ‫thanks to an E Ink marker. 165 00:07:03,180 --> 00:07:06,210 ‫And the data will be loaded onto an S3 bucket, 166 00:07:06,210 --> 00:07:08,430 ‫and then the Snowball will be completely wiped 167 00:07:08,430 --> 00:07:11,100 ‫according to the highest security measures. 168 00:07:11,100 --> 00:07:12,840 ‫So that's for the data migration, 169 00:07:12,840 --> 00:07:14,790 ‫and that was originally, 170 00:07:14,790 --> 00:07:18,000 ‫one and the only use case for Snowball devices. 171 00:07:18,000 --> 00:07:20,490 ‫But the second use case now for the Snow family 172 00:07:20,490 --> 00:07:22,227 ‫is called edge computing. 173 00:07:22,227 --> 00:07:25,290 ‫And so edge computing is when you process data 174 00:07:25,290 --> 00:07:28,860 ‫while it's being created at an edge location. 175 00:07:28,860 --> 00:07:30,630 ‫So what is an edge location? 176 00:07:30,630 --> 00:07:32,460 ‫Well, an edge location is anything 177 00:07:32,460 --> 00:07:34,350 ‫that really doesn't have internet 178 00:07:34,350 --> 00:07:36,540 ‫or that is far away from the cloud. 179 00:07:36,540 --> 00:07:39,060 ‫So for example, if you have a truck on the road 180 00:07:39,060 --> 00:07:40,770 ‫or if you have a ship on the sea 181 00:07:40,770 --> 00:07:42,780 ‫or a mining station on the ground, 182 00:07:42,780 --> 00:07:45,150 ‫all these things can be called edge locations, 183 00:07:45,150 --> 00:07:47,010 ‫because they can produce data 184 00:07:47,010 --> 00:07:50,130 ‫but they may not necessarily have internet connectivity. 185 00:07:50,130 --> 00:07:52,230 ‫So either limited connectivity 186 00:07:52,230 --> 00:07:56,550 ‫or no internet access or no access to computing power. 187 00:07:56,550 --> 00:07:58,800 ‫And so you may still want to run computation, 188 00:07:58,800 --> 00:08:01,380 ‫data processing at these locations. 189 00:08:01,380 --> 00:08:03,810 ‫And for this, we need edge computing. 190 00:08:03,810 --> 00:08:05,250 ‫And so to do edge computing, 191 00:08:05,250 --> 00:08:08,400 ‫we can order a Snowball edge device or a Snowcone, 192 00:08:08,400 --> 00:08:11,790 ‫and have it embedded into these edge locations, 193 00:08:11,790 --> 00:08:13,860 ‫and start doing edge computing. 194 00:08:13,860 --> 00:08:17,130 ‫So the use cases of edge computing is to pre-process data, 195 00:08:17,130 --> 00:08:18,510 ‫do machine learning at the edge, 196 00:08:18,510 --> 00:08:20,640 ‫so without it going back to the cloud, 197 00:08:20,640 --> 00:08:22,770 ‫transcode media streams in advance, 198 00:08:22,770 --> 00:08:24,480 ‫and eventually if need be, 199 00:08:24,480 --> 00:08:27,690 ‫if you need to transfer the data back into AWS, 200 00:08:27,690 --> 00:08:29,520 ‫you can ship back the device 201 00:08:29,520 --> 00:08:31,290 ‫for your Snowcone or your Snowball Edge. 202 00:08:31,290 --> 00:08:33,690 ‫So really you start processing the data very, very close 203 00:08:33,690 --> 00:08:34,980 ‫to where it's being created, 204 00:08:34,980 --> 00:08:38,160 ‫and then you ship it back to AWS. 205 00:08:38,160 --> 00:08:39,990 ‫So for edge computing, what do we have? 206 00:08:39,990 --> 00:08:41,220 ‫We have the Snowcone, 207 00:08:41,220 --> 00:08:44,430 ‫and it comes with two CPUs, four gigabytes of memory, 208 00:08:44,430 --> 00:08:47,280 ‫has wired or wireless access, so wifi, 209 00:08:47,280 --> 00:08:51,840 ‫it's powered by USB-C or an optional battery. 210 00:08:51,840 --> 00:08:54,390 ‫Then for Snowball Edge, it's compute optimized. 211 00:08:54,390 --> 00:08:55,980 ‫So we have the two different flavors again. 212 00:08:55,980 --> 00:08:59,175 ‫So the Compute Optimized one has 104 vCPUs, 213 00:08:59,175 --> 00:09:01,920 ‫416 gigabytes of RAM, 214 00:09:01,920 --> 00:09:03,300 ‫and an optional GPU 215 00:09:03,300 --> 00:09:06,990 ‫if you wanted to do video processing or machine learning. 216 00:09:06,990 --> 00:09:10,530 ‫We have up to 42 terabytes of usable storage. 217 00:09:10,530 --> 00:09:12,360 ‫And for the Snowball edge that is Storage Optimized, 218 00:09:12,360 --> 00:09:14,310 ‫we have 40 vCPU, 80 GB of RAM 219 00:09:14,310 --> 00:09:17,520 ‫and 80 TB of storage. 220 00:09:17,520 --> 00:09:20,970 ‫We have object storage clustering available for the storage. 221 00:09:20,970 --> 00:09:23,010 ‫And all of these devices 222 00:09:23,010 --> 00:09:26,130 ‫can run either EC2 instances within them 223 00:09:26,130 --> 00:09:27,420 ‫or Lambda functions 224 00:09:27,420 --> 00:09:31,110 ‫using the service called AWS IoT Greengrass. 225 00:09:31,110 --> 00:09:32,730 ‫And if you would do edge computing, 226 00:09:32,730 --> 00:09:34,950 ‫then you may want to have these devices 227 00:09:34,950 --> 00:09:36,660 ‫in your facilities, in your trucks, 228 00:09:36,660 --> 00:09:38,940 ‫in your boats for a very long time. 229 00:09:38,940 --> 00:09:41,610 ‫And so therefore you have long-term deployment options 230 00:09:41,610 --> 00:09:43,020 ‫where you can get a discounted pricing 231 00:09:43,020 --> 00:09:47,583 ‫if you borrow these devices for one or three years. 232 00:09:49,260 --> 00:09:51,390 ‫Finally, for the Snow family, 233 00:09:51,390 --> 00:09:53,820 ‫there is Snow, sorry, OpsHub. 234 00:09:53,820 --> 00:09:56,670 ‫So historically, when we're using these devices 235 00:09:56,670 --> 00:09:59,730 ‫you needed a CLI, so command line interface tool, 236 00:09:59,730 --> 00:10:02,220 ‫to deal with them, and it was very, very difficult. 237 00:10:02,220 --> 00:10:04,020 ‫And so AWS recognizes that. 238 00:10:04,020 --> 00:10:06,750 ‫And so they've created OpsHub, which is a software 239 00:10:06,750 --> 00:10:08,730 ‫that you install on your computer or laptop. 240 00:10:08,730 --> 00:10:10,080 ‫So it's not something you use on the cloud. 241 00:10:10,080 --> 00:10:12,690 ‫It's something you have to download on your computer. 242 00:10:12,690 --> 00:10:13,800 ‫And then once it's connected, 243 00:10:13,800 --> 00:10:16,500 ‫it's going to give you a graphical interface 244 00:10:16,500 --> 00:10:18,810 ‫to connect to your Snow devices, 245 00:10:18,810 --> 00:10:22,350 ‫and configure them and use them, which is very, very handy. 246 00:10:22,350 --> 00:10:23,183 ‫So this is obviously, 247 00:10:23,183 --> 00:10:26,760 ‫to do unlocking and configuring single or cluster devices, 248 00:10:26,760 --> 00:10:29,610 ‫transferring files, launching and managing instances, 249 00:10:29,610 --> 00:10:32,760 ‫so EC2 instances running on Snow family devices, 250 00:10:32,760 --> 00:10:34,440 ‫monitor device metrics, 251 00:10:34,440 --> 00:10:37,320 ‫and launch compatible AWS services on your devices. 252 00:10:37,320 --> 00:10:38,970 ‫For example, EC2 instances, 253 00:10:38,970 --> 00:10:42,630 ‫use DataSync or a network file system. 254 00:10:42,630 --> 00:10:44,160 ‫So that's it for the Snow family. 255 00:10:44,160 --> 00:10:47,003 ‫I hope you liked it, and I will see you in the next lecture.