1 00:00:00,260 --> 00:00:03,270 So we've seen a way to attach a network drive 2 00:00:03,270 --> 00:00:04,838 onto our EC2 Instances 3 00:00:04,838 --> 00:00:07,440 but they have limited performance 4 00:00:07,440 --> 00:00:10,045 and I said with cause because it's a really good performance 5 00:00:10,045 --> 00:00:13,670 but sometimes you want something even higher performance 6 00:00:13,670 --> 00:00:16,290 and that is going to be a hardware disk 7 00:00:16,290 --> 00:00:18,329 attached onto you EC2 Instance. 8 00:00:18,329 --> 00:00:21,499 So the EC2 Instance is a virtual machine 9 00:00:21,499 --> 00:00:25,370 but it is obviously attached to a real hardware server. 10 00:00:25,370 --> 00:00:28,500 And some of these servers do have disk space 11 00:00:28,500 --> 00:00:30,750 that is attached directly you know, 12 00:00:30,750 --> 00:00:33,730 with a physical connection onto the server. 13 00:00:33,730 --> 00:00:36,970 And so a special type of EC2 Instance can leverage 14 00:00:36,970 --> 00:00:39,136 something called an EC2 Instance Store, 15 00:00:39,136 --> 00:00:41,095 which is the name of the hardware, 16 00:00:41,095 --> 00:00:44,603 the hard drive attached to the physical server. 17 00:00:44,603 --> 00:00:46,641 So what we do with EC2 Instance Store, 18 00:00:46,641 --> 00:00:49,510 we use them for better I/O performance. 19 00:00:49,510 --> 00:00:52,240 We also make sure that they have good through put 20 00:00:52,240 --> 00:00:53,850 and so on, so they're a great choice 21 00:00:53,850 --> 00:00:56,880 when you want to have extremely high disk performance. 22 00:00:56,880 --> 00:00:59,766 But the caveat is that if you stop 23 00:00:59,766 --> 00:01:02,800 or you terminate your EC2 Instance, 24 00:01:02,800 --> 00:01:04,550 that has an Instance Store, 25 00:01:04,550 --> 00:01:06,910 then the storage will be lost. 26 00:01:06,910 --> 00:01:09,149 And therefore it's called an ephemeral storage 27 00:01:09,149 --> 00:01:11,380 so that means that the EC2 Instance Store 28 00:01:11,380 --> 00:01:14,930 can now be used as a durable long term place 29 00:01:14,930 --> 00:01:16,430 to store your data. 30 00:01:16,430 --> 00:01:18,740 So what is a good use case for it then? 31 00:01:18,740 --> 00:01:20,580 Well if you have a buffer, a cache, 32 00:01:20,580 --> 00:01:23,621 you want to have scratch data or temporary content, 33 00:01:23,621 --> 00:01:26,870 this would be a great place to do these things 34 00:01:26,870 --> 00:01:28,110 but not for long term storage. 35 00:01:28,110 --> 00:01:30,070 For long term storage, EBS for example 36 00:01:30,070 --> 00:01:31,810 is a great use case. 37 00:01:31,810 --> 00:01:34,349 Finally, in case the on the line server 38 00:01:34,349 --> 00:01:37,517 of the EC2 Instance does fail, 39 00:01:37,517 --> 00:01:39,530 then you'll risk to have a dear loss 40 00:01:39,530 --> 00:01:41,950 because the hardware attached to the EC2 Instance 41 00:01:41,950 --> 00:01:43,340 will fail as well. 42 00:01:43,340 --> 00:01:46,322 So if you do decide to use an EC2 Instance Store, 43 00:01:46,322 --> 00:01:48,373 then it is your entire responsibility 44 00:01:48,373 --> 00:01:50,511 to make sure that you back it up 45 00:01:50,511 --> 00:01:54,015 and that you replicate it correctly based on your needs. 46 00:01:54,015 --> 00:01:55,880 So what I mean by better performance, 47 00:01:55,880 --> 00:01:58,400 this is just an example to illustrate it, 48 00:01:58,400 --> 00:01:59,233 don't need to know it. 49 00:01:59,233 --> 00:02:01,130 But if you look at for example, 50 00:02:01,130 --> 00:02:03,764 the Instance size of I3 dot something, 51 00:02:03,764 --> 00:02:07,110 there is an Instance Store attached 52 00:02:07,110 --> 00:02:09,070 to these kinds of instances 53 00:02:09,070 --> 00:02:10,990 and if you look at the Read IOPS 54 00:02:10,990 --> 00:02:11,823 and the Write IOPS 55 00:02:11,823 --> 00:02:13,770 which correspond to how many I/O operations 56 00:02:13,770 --> 00:02:15,120 we can do per second. 57 00:02:15,120 --> 00:02:17,980 Then you can see at some of these random Read IOPS 58 00:02:17,980 --> 00:02:21,575 and Write IOPS can reach 3.3 million or 1.4 million, 59 00:02:21,575 --> 00:02:24,590 for the most performant one. 60 00:02:24,590 --> 00:02:26,200 And to put this in comparison 61 00:02:26,200 --> 00:02:28,820 with an EBS volume of type BP2 for example, 62 00:02:28,820 --> 00:02:31,407 you can reach thirty two thousand IOPS. 63 00:02:31,407 --> 00:02:33,450 So this is a lot more. 64 00:02:33,450 --> 00:02:34,931 But again, it's just to illustrate my point 65 00:02:34,931 --> 00:02:37,960 from an exam perspective anytime you see very 66 00:02:37,960 --> 00:02:40,762 high performance hardware attached volume 67 00:02:40,762 --> 00:02:45,350 for your EC2 Instances, think local EC2 Instance Store. 68 00:02:45,350 --> 00:02:46,183 That's it. 69 00:02:46,183 --> 00:02:47,700 I will see you in the next lecture.