1 00:00:00,120 --> 00:00:01,680 So let's talk about Amazon Aurora 2 00:00:01,680 --> 00:00:02,910 because the exam is starting 3 00:00:02,910 --> 00:00:04,650 to ask you a lot of questions about it. 4 00:00:04,650 --> 00:00:06,780 Now, you don't need deep knowledge on it, 5 00:00:06,780 --> 00:00:08,880 but you need enough high level overview 6 00:00:08,880 --> 00:00:10,350 to understand exactly how it works, 7 00:00:10,350 --> 00:00:11,183 So this is what I'm going 8 00:00:11,183 --> 00:00:12,660 to give you in this lecture. 9 00:00:12,660 --> 00:00:13,860 Aurora is going to be 10 00:00:13,860 --> 00:00:15,870 a proprietary technology from AWS. 11 00:00:15,870 --> 00:00:17,550 It's not open sourced. 12 00:00:17,550 --> 00:00:19,650 But they made it so it's going to be compatible 13 00:00:19,650 --> 00:00:21,510 with Postgres and MySQL. 14 00:00:21,510 --> 00:00:22,980 And basically Aurora database 15 00:00:22,980 --> 00:00:24,390 will have compatible drivers. 16 00:00:24,390 --> 00:00:25,830 That means that if you connect 17 00:00:25,830 --> 00:00:27,330 as if you were connecting to a Postgres 18 00:00:27,330 --> 00:00:28,410 or a MySQL database, 19 00:00:28,410 --> 00:00:29,850 then it will work. 20 00:00:29,850 --> 00:00:30,930 Aurora is very special 21 00:00:30,930 --> 00:00:32,940 and I won't go too deep into the internals, 22 00:00:32,940 --> 00:00:34,530 but they made it cloud optimized. 23 00:00:34,530 --> 00:00:36,360 And by doing a lot of optimization 24 00:00:36,360 --> 00:00:37,530 and smart stuff, 25 00:00:37,530 --> 00:00:39,990 basically they get 5x performance improvements 26 00:00:39,990 --> 00:00:41,490 over MySQL on RDS 27 00:00:41,490 --> 00:00:44,190 or 3x performance of Postgres on RDS. 28 00:00:44,190 --> 00:00:45,840 Not just that, but in many different ways 29 00:00:45,840 --> 00:00:47,820 they also gets more performance improvements. 30 00:00:47,820 --> 00:00:48,653 To me, I watch it. 31 00:00:48,653 --> 00:00:49,486 It's really, really smart, 32 00:00:49,486 --> 00:00:51,360 but I won't go into the details of it. 33 00:00:51,360 --> 00:00:53,460 Now, Aurora storage automatically grows. 34 00:00:53,460 --> 00:00:54,360 And I think this is one 35 00:00:54,360 --> 00:00:56,370 of the main features that is quite awesome. 36 00:00:56,370 --> 00:00:57,630 You start at 10 gigabytes, 37 00:00:57,630 --> 00:01:01,200 but as you put more data into your database 38 00:01:01,200 --> 00:01:04,560 it grows automatically up to 128 terabytes. 39 00:01:04,560 --> 00:01:06,390 Again, this has to do with how to design it, 40 00:01:06,390 --> 00:01:08,340 but the awesome thing is that now as a DB, 41 00:01:08,340 --> 00:01:10,080 as a DB or a SysOps 42 00:01:10,080 --> 00:01:12,690 you don't need to worry about monitoring your disc. 43 00:01:12,690 --> 00:01:15,870 You just know it will grow automatically with time. 44 00:01:15,870 --> 00:01:17,220 Also for the read replicas, 45 00:01:17,220 --> 00:01:19,230 you can have up to 15 replica as well. 46 00:01:19,230 --> 00:01:21,960 MySQL only has five and the replication faster 47 00:01:21,960 --> 00:01:23,850 the way they made it it's much faster. 48 00:01:23,850 --> 00:01:26,070 So overall it's a win. 49 00:01:26,070 --> 00:01:27,570 Now, if you do failover in Aurora 50 00:01:27,570 --> 00:01:29,490 it was going to be instantaneous. 51 00:01:29,490 --> 00:01:30,720 So it's gonna be much faster 52 00:01:30,720 --> 00:01:34,140 than a failover from Multi-AZ on MySQL on RDS. 53 00:01:34,140 --> 00:01:36,030 And because it's cloud native by default 54 00:01:36,030 --> 00:01:38,640 you get high availability. 55 00:01:38,640 --> 00:01:40,680 Now, although the cost is a little bit more than RDS, 56 00:01:40,680 --> 00:01:43,560 about 20% more, it is so much more efficient 57 00:01:43,560 --> 00:01:46,710 that at scale, it makes a lot more sense for savings. 58 00:01:46,710 --> 00:01:49,530 So let's talk about the aspects that are super important 59 00:01:49,530 --> 00:01:51,960 which is high availability and rescaling. 60 00:01:51,960 --> 00:01:53,130 So Aurora is special 61 00:01:53,130 --> 00:01:56,160 because it's going to store six copies of your data 62 00:01:56,160 --> 00:01:59,310 anytime you write anything across 3 AZ. 63 00:01:59,310 --> 00:02:01,920 And so Aurora is made such as it's available. 64 00:02:01,920 --> 00:02:04,890 So it only needs four copies out of six for writes. 65 00:02:04,890 --> 00:02:07,590 So that means that if one AZ is down, you're fine. 66 00:02:07,590 --> 00:02:09,060 And it only means you have three copies 67 00:02:09,060 --> 00:02:11,009 out of six needed for reads. 68 00:02:11,009 --> 00:02:14,070 So again, that means that it's highly available for reads. 69 00:02:14,070 --> 00:02:16,170 There is some kind of self healing process that happens, 70 00:02:16,170 --> 00:02:18,870 which is quite cool, which is that if some data is corrupted 71 00:02:18,870 --> 00:02:21,420 or bad, then it does self healing 72 00:02:21,420 --> 00:02:23,400 with peer-to-peer replication in the backend. 73 00:02:23,400 --> 00:02:24,330 And it's quite cool. 74 00:02:24,330 --> 00:02:26,250 And you don't rely on just one volume. 75 00:02:26,250 --> 00:02:28,027 You rely on hundreds of volumes, again, 76 00:02:28,027 --> 00:02:29,160 that's not something for you to manage. 77 00:02:29,160 --> 00:02:30,390 It happens in the backend, 78 00:02:30,390 --> 00:02:33,420 but that means that you just reduced the risk by so much. 79 00:02:33,420 --> 00:02:35,790 So if you look at it from a diagram perspective 80 00:02:35,790 --> 00:02:38,640 you have three AZ and you have a shared storage volume, 81 00:02:38,640 --> 00:02:39,960 but it's a logical volume. 82 00:02:39,960 --> 00:02:42,510 And it does replication, self healing, and auto expanding, 83 00:02:42,510 --> 00:02:44,100 which is a lot of features. 84 00:02:44,100 --> 00:02:47,040 So if you were to write some data, maybe blue data, 85 00:02:47,040 --> 00:02:50,280 you'll see the six copies of it in three different AZ. 86 00:02:50,280 --> 00:02:51,750 Then if you write some orange data, 87 00:02:51,750 --> 00:02:53,907 again six copies of it in different AZ. 88 00:02:53,907 --> 00:02:55,830 And then as you write more and more data, 89 00:02:55,830 --> 00:02:58,350 it's basically going to go again six copies of it 90 00:02:58,350 --> 00:02:59,550 in three different AZ. 91 00:02:59,550 --> 00:03:01,410 The cool thing is that it goes on different volumes 92 00:03:01,410 --> 00:03:03,960 and it's striped and it works really, really well. 93 00:03:03,960 --> 00:03:05,490 Now we need to know about storage, 94 00:03:05,490 --> 00:03:08,190 and that's it, but you don't actually interface 95 00:03:08,190 --> 00:03:09,023 with the storage. 96 00:03:09,023 --> 00:03:10,470 It's just a design that Amazon made. 97 00:03:10,470 --> 00:03:12,030 And I want to give it to you as well 98 00:03:12,030 --> 00:03:14,910 so you understand what Aurora takes. 99 00:03:14,910 --> 00:03:18,900 Now Aurora is like Multi-AZ for RDS. 100 00:03:18,900 --> 00:03:21,360 Basically there's only one instance that takes writes. 101 00:03:21,360 --> 00:03:23,370 So there is a master in Aurora 102 00:03:23,370 --> 00:03:24,990 and that's where we'll take writes. 103 00:03:24,990 --> 00:03:27,750 And then if the master doesn't work, 104 00:03:27,750 --> 00:03:31,200 the failover will happen in less than 30 seconds on average. 105 00:03:31,200 --> 00:03:33,240 So it's really, really quick failover. 106 00:03:33,240 --> 00:03:34,650 And on top of the master, 107 00:03:34,650 --> 00:03:38,700 you can have up to 15 read replicas, all serving reads. 108 00:03:38,700 --> 00:03:40,140 So you can have a lot of them. 109 00:03:40,140 --> 00:03:42,210 And this is how you scale your read workload. 110 00:03:42,210 --> 00:03:45,060 And so any of these read replicas can become the master 111 00:03:45,060 --> 00:03:46,470 in case the master fails. 112 00:03:46,470 --> 00:03:49,080 So it's quite different from how RDS works, 113 00:03:49,080 --> 00:03:51,780 but by default, you only have one master. 114 00:03:51,780 --> 00:03:53,370 The cool thing about these read replicas is 115 00:03:53,370 --> 00:03:56,160 that it supports cross region replication. 116 00:03:56,160 --> 00:03:58,650 So if you look at Aurora on the right hand side 117 00:03:58,650 --> 00:04:00,480 of diagram, this is what you should remember. 118 00:04:00,480 --> 00:04:02,490 One master, multiple root replicas, 119 00:04:02,490 --> 00:04:05,700 and the storage is gonna be replicated, self healing, 120 00:04:05,700 --> 00:04:08,580 auto expanding little blocks by little blocks. 121 00:04:08,580 --> 00:04:11,820 Now let's have a look at how Aurora is as a cluster. 122 00:04:11,820 --> 00:04:14,400 So this is more around how Aurora works. 123 00:04:14,400 --> 00:04:15,233 When you have clients 124 00:04:15,233 --> 00:04:18,060 how do you interface with all these instances? 125 00:04:18,060 --> 00:04:20,100 So, as we said, we have a shared storage volume 126 00:04:20,100 --> 00:04:24,540 which is auto expanding from 10 gigabytes to 128 gigabytes. 127 00:04:24,540 --> 00:04:26,850 Your master is the only thing that will write 128 00:04:26,850 --> 00:04:28,050 to your storage. 129 00:04:28,050 --> 00:04:30,540 And because the master can change and fail over, 130 00:04:30,540 --> 00:04:33,870 what Aurora provides you is what's called a writer endpoint. 131 00:04:33,870 --> 00:04:35,970 So it's a DNS name, a writer endpoint, 132 00:04:35,970 --> 00:04:37,560 and it's always pointing to the master. 133 00:04:37,560 --> 00:04:39,360 So even if the master fails over, 134 00:04:39,360 --> 00:04:41,610 your clients still talks to the writer endpoint 135 00:04:41,610 --> 00:04:44,850 and is automatically redirected to the right instance. 136 00:04:44,850 --> 00:04:46,050 Now, as I said before, 137 00:04:46,050 --> 00:04:48,060 you also have a lot of read replicas. 138 00:04:48,060 --> 00:04:49,170 What I didn't say is 139 00:04:49,170 --> 00:04:51,000 that they can have auto scaling 140 00:04:51,000 --> 00:04:52,410 on top of these read replicas. 141 00:04:52,410 --> 00:04:54,990 So you can have one up to 15 read replicas 142 00:04:54,990 --> 00:04:57,420 and you can set up auto scaling such as you always 143 00:04:57,420 --> 00:05:00,150 have the right number of read replicas. 144 00:05:00,150 --> 00:05:01,470 Now because you have auto scaling, 145 00:05:01,470 --> 00:05:03,450 it can be really, really hard for your applications 146 00:05:03,450 --> 00:05:06,090 to keep track of where are your read replicas? 147 00:05:06,090 --> 00:05:06,923 What's the URL? 148 00:05:06,923 --> 00:05:08,370 How do I connect to them? 149 00:05:08,370 --> 00:05:10,470 So for it, you have to remember this absolutely 150 00:05:10,470 --> 00:05:11,700 for going for the exam. 151 00:05:11,700 --> 00:05:14,070 There is something called a reader endpoint. 152 00:05:14,070 --> 00:05:16,140 And a reader endpoint has the exact same feature 153 00:05:16,140 --> 00:05:17,250 as a writer endpoint. 154 00:05:17,250 --> 00:05:19,110 It helps with connection load balancing 155 00:05:19,110 --> 00:05:22,620 and it connects automatically to all the read replicas. 156 00:05:22,620 --> 00:05:26,010 So any time the client connects to the reader endpoint, 157 00:05:26,010 --> 00:05:28,440 it will get connected to one of the read replicas, 158 00:05:28,440 --> 00:05:31,290 and there will be load balancing done this way. 159 00:05:31,290 --> 00:05:34,320 Make sure, just notice that the load balancing 160 00:05:34,320 --> 00:05:37,440 happens at the connection level, not the statement level. 161 00:05:37,440 --> 00:05:39,540 So this is how it works for Aurora. 162 00:05:39,540 --> 00:05:41,970 Remember writer endpoint, reader endpoint. 163 00:05:41,970 --> 00:05:43,080 Remember auto scaling. 164 00:05:43,080 --> 00:05:45,480 Remember shared storage volume that auto expands. 165 00:05:45,480 --> 00:05:47,550 Remember this diagram, because once you get it, 166 00:05:47,550 --> 00:05:49,770 you understand how Aurora works. 167 00:05:49,770 --> 00:05:51,840 Now, if we go deep into the feature, 168 00:05:51,840 --> 00:05:53,490 you get a lot of things I already told you. 169 00:05:53,490 --> 00:05:55,800 Automatic failover, backup and recovery, 170 00:05:55,800 --> 00:05:58,470 isolation and security, industry compliance, 171 00:05:58,470 --> 00:06:00,840 push-button scaling by auto scaling, 172 00:06:00,840 --> 00:06:02,400 automated patching with zero downtime, 173 00:06:02,400 --> 00:06:05,130 so it's kind of cool they do in the backend, 174 00:06:05,130 --> 00:06:07,230 advanced monitoring, routine maintenance. 175 00:06:07,230 --> 00:06:08,820 So all these things are handled for you. 176 00:06:08,820 --> 00:06:10,920 And you also get this feature called backtrack, 177 00:06:10,920 --> 00:06:12,960 which is giving you the ability to restore data 178 00:06:12,960 --> 00:06:14,490 at any point of time. 179 00:06:14,490 --> 00:06:16,290 And it actually doesn't rely on backups. 180 00:06:16,290 --> 00:06:17,670 It relies on something different. 181 00:06:17,670 --> 00:06:19,440 But you can always say, "I want to go back 182 00:06:19,440 --> 00:06:20,550 to yesterday at 4:00 PM," 183 00:06:20,550 --> 00:06:22,560 and you say, "Oh no, actually I wanted to do 184 00:06:22,560 --> 00:06:23,970 yesterday at 5:00 PM." 185 00:06:23,970 --> 00:06:26,910 And it will work as well, which is super, super neat. 186 00:06:26,910 --> 00:06:28,050 So that's it for Aurora. 187 00:06:28,050 --> 00:06:30,000 And I will see you in the next lecture.