1 00:00:00,480 --> 00:00:02,830 So now let's talk about EBS volumes. 2 00:00:02,830 --> 00:00:04,270 And there are different volume types. 3 00:00:04,270 --> 00:00:06,350 They come in six different types today. 4 00:00:06,350 --> 00:00:08,740 And we can group them in several categories. 5 00:00:08,740 --> 00:00:11,140 The first one is, gp2/gp3. 6 00:00:11,140 --> 00:00:13,390 Which has a general purpose SSD volume 7 00:00:13,390 --> 00:00:15,160 that will balance price and performance 8 00:00:15,160 --> 00:00:17,000 for a wide variety of workloads. 9 00:00:17,000 --> 00:00:17,840 And this is something we've been 10 00:00:17,840 --> 00:00:19,860 using so far in this course. 11 00:00:19,860 --> 00:00:21,950 Then we have io1 and io2. 12 00:00:21,950 --> 00:00:24,080 They're highest performance SSD volumes 13 00:00:24,080 --> 00:00:25,560 that will be used for mission critical 14 00:00:25,560 --> 00:00:28,570 low latency and high throughput workloads. 15 00:00:28,570 --> 00:00:30,380 Then we have st 1 volumes, 16 00:00:30,380 --> 00:00:32,360 and they're low-cost HDD volume designed 17 00:00:32,360 --> 00:00:35,470 for frequently accessed throughput intensive workloads. 18 00:00:35,470 --> 00:00:36,670 We have sc 1 volume 19 00:00:36,670 --> 00:00:39,450 which are going to be the lowest cost HDD volume, 20 00:00:39,450 --> 00:00:40,500 and it will be designed 21 00:00:40,500 --> 00:00:43,020 for less frequently accessed workloads. 22 00:00:43,020 --> 00:00:45,410 Now, how do you define an EBS volume? 23 00:00:45,410 --> 00:00:47,050 Well, you can have several factors. 24 00:00:47,050 --> 00:00:50,620 For example size, throughputs and IOPS. 25 00:00:50,620 --> 00:00:54,040 Which means I/O operations per second. 26 00:00:54,040 --> 00:00:55,510 And obviously when in doubt, 27 00:00:55,510 --> 00:00:58,010 always consult the documentation. 28 00:00:58,010 --> 00:01:00,513 Now for your EC2 instances, 29 00:01:00,513 --> 00:01:03,840 only gp2 and gp3 and io1 and io2 30 00:01:03,840 --> 00:01:05,540 can be used as boot volumes. 31 00:01:05,540 --> 00:01:09,670 That means where the root OS is going to be running. 32 00:01:09,670 --> 00:01:13,930 So now let's go do a deeper dive on to gp2/gp3, io1/io2 33 00:01:13,930 --> 00:01:14,763 and the other ones. 34 00:01:14,763 --> 00:01:17,740 But gp2, the general purpose and the provision IOPS 35 00:01:17,740 --> 00:01:20,010 are going to be the most important for your exam. 36 00:01:20,010 --> 00:01:23,720 So gp2 is cost-effective storage with low latency 37 00:01:23,720 --> 00:01:25,510 and you can use it for system boot volumes, 38 00:01:25,510 --> 00:01:27,890 virtual desktop, development and test environments. 39 00:01:27,890 --> 00:01:31,150 The size can vary between one gigabyte to 16 terabytes. 40 00:01:31,150 --> 00:01:34,037 And so we have differences between gp2 and gp3. 41 00:01:34,037 --> 00:01:37,180 Gp3 is the newer generation of volumes. 42 00:01:37,180 --> 00:01:40,050 So gp3 gives you a baseline of 3000 IOPS 43 00:01:40,050 --> 00:01:43,080 and a throughput of 125 megabytes per second. 44 00:01:43,080 --> 00:01:46,276 And then we can increase IOPS up to 16,000 45 00:01:46,276 --> 00:01:47,710 and throughput up to 46 00:01:47,710 --> 00:01:50,580 1000 megabytes per second independently. 47 00:01:50,580 --> 00:01:52,470 Okay, so they're not linked. 48 00:01:52,470 --> 00:01:54,423 For gp2, this is the older version. 49 00:01:54,423 --> 00:01:58,360 They're small gp2 volumes they can burst up to 3000 IOPS. 50 00:01:58,360 --> 00:02:01,280 And then the size of the volume and the IOPS are linked. 51 00:02:01,280 --> 00:02:03,970 That means that if you increase the IOPS, 52 00:02:03,970 --> 00:02:06,160 if you increase the number of gigabytes on your volume, 53 00:02:06,160 --> 00:02:11,070 then you're going to get three more IOPS up to 16,000 IOPS. 54 00:02:11,070 --> 00:02:15,410 And so that means that if you have 5,333 or 34 gigabytes 55 00:02:15,410 --> 00:02:18,110 then you will have 16000 IOPS and you've maxed it out. 56 00:02:18,110 --> 00:02:20,010 So what do you remember from this slide? 57 00:02:20,010 --> 00:02:22,300 You remember, that in gp2/gp3, 58 00:02:22,300 --> 00:02:25,190 this is for cost-effective storage latency. 59 00:02:25,190 --> 00:02:28,160 And in gp3 you can independently set the IOPS 60 00:02:28,160 --> 00:02:29,190 and the throughputs, 61 00:02:29,190 --> 00:02:31,960 whereas for gp2, they're linked together. 62 00:02:31,960 --> 00:02:33,740 Now the other types of volume that will come 63 00:02:33,740 --> 00:02:36,070 up in the exam is provisioned IOPS. 64 00:02:36,070 --> 00:02:38,540 And this is going to be a very good use case 65 00:02:38,540 --> 00:02:41,090 when you have a critical business application 66 00:02:41,090 --> 00:02:43,820 that needs to sustain the IOPS performance 67 00:02:43,820 --> 00:02:46,910 or applications that need more than 16,000 IOPS. 68 00:02:46,910 --> 00:02:49,180 So in general you will have a use case 69 00:02:49,180 --> 00:02:50,660 where you have a database workload. 70 00:02:50,660 --> 00:02:53,200 For example, something that really uses the storage, 71 00:02:53,200 --> 00:02:54,790 and it is very sensitive 72 00:02:54,790 --> 00:02:57,020 to storage performance and consistency. 73 00:02:57,020 --> 00:03:00,440 In that case, switching from a gp2 or gp3 volume 74 00:03:00,440 --> 00:03:03,990 to io1 or io2 volume is going to be the answer. 75 00:03:03,990 --> 00:03:07,090 So if we consider io1/io2 you choose the newer generation. 76 00:03:07,090 --> 00:03:09,890 They can be between four and 16 terabytes. 77 00:03:09,890 --> 00:03:11,200 The max IOPS you're going to get 78 00:03:11,200 --> 00:03:14,650 is 60,000 IOPS for Nitro EC2 instances. 79 00:03:14,650 --> 00:03:17,160 So remember, Nitro Ec2 instances, gives you, 80 00:03:17,160 --> 00:03:21,030 Nitro gives you access to a higher level of IOPS. 81 00:03:21,030 --> 00:03:23,130 And if you don't have a Nitro Ec2 instance, 82 00:03:23,130 --> 00:03:27,700 then you're going to get 30,000, 32,000 IOPS maximum. 83 00:03:27,700 --> 00:03:31,400 And with io1 and io2 you can increase the provision IOPS 84 00:03:31,400 --> 00:03:34,460 independently from the storage size, just like gp3. 85 00:03:34,460 --> 00:03:36,310 And why do we use io2? 86 00:03:36,310 --> 00:03:38,620 Well, it has more durability and more IOPS 87 00:03:38,620 --> 00:03:40,940 per gigabytes at the same price of io1. 88 00:03:40,940 --> 00:03:44,110 So today it makes just sense to use io2. 89 00:03:44,110 --> 00:03:47,140 And there is a preview of io2 Block Express 90 00:03:47,140 --> 00:03:51,400 which is between four gigabytes and 64 terabytes. 91 00:03:51,400 --> 00:03:54,920 And this is an even higher performance type of volumes. 92 00:03:54,920 --> 00:03:57,116 This gives you sub millisecond latency 93 00:03:57,116 --> 00:04:01,990 and you're going to get 256,000 IOPS maximum 94 00:04:01,990 --> 00:04:05,520 with a ratio of IOPS per gigabyte of one thousand per one. 95 00:04:05,520 --> 00:04:07,970 And finally we'll see this in the next lecture, 96 00:04:07,970 --> 00:04:10,760 The EBS volume types of provision IOPS 97 00:04:10,760 --> 00:04:13,340 supports EBS multi attach. 98 00:04:13,340 --> 00:04:17,620 So now let's quickly touch upon st1 and sc1. 99 00:04:17,620 --> 00:04:19,103 So they can not be a boot volumes. 100 00:04:19,103 --> 00:04:21,300 This is going to be for the previous types of volumes. 101 00:04:21,300 --> 00:04:24,210 You can get the size up to 16 terabytes, 102 00:04:24,210 --> 00:04:26,750 and we get two kinds of volumes here. 103 00:04:26,750 --> 00:04:29,870 We have the throughput optimized HDD, so st 1. 104 00:04:29,870 --> 00:04:31,010 which is great for big data, 105 00:04:31,010 --> 00:04:33,340 data warehousing, and log processing. 106 00:04:33,340 --> 00:04:34,640 Which gives you a max throughput 107 00:04:34,640 --> 00:04:38,470 of 500 megabytes per second and a max IOPS of 500. 108 00:04:38,470 --> 00:04:40,850 And then for cold HDD, so sc1. 109 00:04:40,850 --> 00:04:42,410 This is for archive data. 110 00:04:42,410 --> 00:04:45,030 So data that is infrequently accessed. 111 00:04:45,030 --> 00:04:46,253 And this is when you would use it, 112 00:04:46,253 --> 00:04:48,720 when you need to have the lowest possible cost. 113 00:04:48,720 --> 00:04:51,970 Here the max throughput is 250 megabytes per second 114 00:04:51,970 --> 00:04:54,460 and the max IOPS is 250 as well. 115 00:04:54,460 --> 00:04:55,430 Now you don't have to remember 116 00:04:55,430 --> 00:04:57,260 all these details going into the exam. 117 00:04:57,260 --> 00:04:58,560 You just need to understand the differences 118 00:04:58,560 --> 00:05:00,670 at a high level of all these volumes. 119 00:05:00,670 --> 00:05:02,355 So the general purpose HHD 120 00:05:02,355 --> 00:05:04,400 versus the provision IOPS HDD 121 00:05:04,400 --> 00:05:05,660 in case you need the database 122 00:05:05,660 --> 00:05:08,370 versus some st 1 in case 123 00:05:08,370 --> 00:05:11,490 you need to have high throughput and lowest cost. 124 00:05:11,490 --> 00:05:13,010 And so you can find a summary 125 00:05:13,010 --> 00:05:15,947 of all I said in this link right here, okay? 126 00:05:15,947 --> 00:05:18,000 And which is a screenshot I just took you. 127 00:05:18,000 --> 00:05:20,300 So don't need to remember the details again 128 00:05:20,300 --> 00:05:24,020 but remember that if you want to get also over 32,000 IOPS, 129 00:05:24,020 --> 00:05:27,730 you need Ec2 Nitro with io1 or io2, 130 00:05:27,730 --> 00:05:29,500 and that will be it for this lecture. 131 00:05:29,500 --> 00:05:30,370 So I hope you liked it. 132 00:05:30,370 --> 00:05:32,320 And I will see you in the next lecture.