1
00:00:00,210 --> 00:00:01,780
‫What is going on, guys?

2
00:00:01,800 --> 00:00:04,790
‫My name is Hussein, and this is the third episode of Suffragette.

3
00:00:04,980 --> 00:00:07,520
‫Today's topic is distributor transaction.

4
00:00:07,530 --> 00:00:12,270
‫Some of you ask me why this what is this thing distributed?

5
00:00:12,720 --> 00:00:17,880
‫And to be honest, when I heard that when I heard that question, what is distributing a distribution

6
00:00:17,880 --> 00:00:18,490
‫transaction?

7
00:00:18,510 --> 00:00:20,010
‫What is the saga thing?

8
00:00:20,010 --> 00:00:21,600
‫And I didn't know anything about it.

9
00:00:22,320 --> 00:00:29,160
‫I read some articles talking about distribution transaction with within the context of Google Atomic

10
00:00:29,160 --> 00:00:33,870
‫Clock, and I know I exist and this is then, but I didn't know why.

11
00:00:34,440 --> 00:00:40,550
‫In the vernacular, why does this technology exist and why do we need distribution transaction?

12
00:00:40,560 --> 00:00:41,610
‫What other solutions?

13
00:00:41,630 --> 00:00:44,370
‫So I spent some time doing some research.

14
00:00:44,940 --> 00:00:51,640
‫I'm going to reference all the videos I wash all the articles I read to kind of summarize this for you.

15
00:00:51,660 --> 00:00:55,520
‫So and this video, I'm going to talk about distribution transactions.

16
00:00:55,830 --> 00:00:57,030
‫Why does it exist?

17
00:00:57,030 --> 00:01:02,670
‫Because I like all of us to start that any technology in software engineering exist for a reason.

18
00:01:02,940 --> 00:01:05,610
‫And if it exists, it solves the problem.

19
00:01:05,610 --> 00:01:06,840
‫What problems does solve?

20
00:01:06,960 --> 00:01:07,290
‫Right.

21
00:01:07,620 --> 00:01:08,880
‫And maybe.

22
00:01:09,920 --> 00:01:15,840
‫That problem could happen to certain sector of engineers, not everybody.

23
00:01:16,370 --> 00:01:18,080
‫That's why I'd like to address this problem.

24
00:01:18,130 --> 00:01:20,480
‫Like, do you really need distributer transaction?

25
00:01:20,510 --> 00:01:22,220
‫Well, what's your problem?

26
00:01:22,220 --> 00:01:23,900
‫Like, that's that's why I'd like to do it.

27
00:01:24,530 --> 00:01:31,220
‫And then I'm going to talk about the different solutions to distribution transactions, because there's

28
00:01:31,460 --> 00:01:32,990
‫that's the beauty of social engineering.

29
00:01:33,300 --> 00:01:36,020
‫There is no one way to solve a problem.

30
00:01:36,050 --> 00:01:37,400
‫There are many, many, many ways.

31
00:01:37,870 --> 00:01:43,180
‫That's what makes us really, really a creative endeavor field.

32
00:01:43,190 --> 00:01:47,520
‫If you're a software engineer, you're and you come up with solutions, you're a create.

33
00:01:47,570 --> 00:01:49,580
‫You're like an artist, man.

34
00:01:49,730 --> 00:01:55,490
‫Just, yeah, you're an artist because coming up with creative solutions to solve our engineering problems,

35
00:01:55,490 --> 00:01:58,490
‫it's it's an art, whether you like it or not.

36
00:01:58,620 --> 00:02:02,750
‫So I respect anyone whether the solution is bad or good.

37
00:02:03,290 --> 00:02:04,580
‫It's an art by itself.

38
00:02:04,610 --> 00:02:10,550
‫You don't look at a painting and say, oh, that's that's not that's that's not the right painting or

39
00:02:11,090 --> 00:02:12,440
‫that's a wrong painting.

40
00:02:12,440 --> 00:02:13,600
‫You know, you can say that, right?

41
00:02:13,700 --> 00:02:15,290
‫It's just it's an expression.

42
00:02:15,290 --> 00:02:15,980
‫And I like it.

43
00:02:17,090 --> 00:02:18,710
‫That being said, let's just jump into it.

44
00:02:19,250 --> 00:02:19,520
‫All right.

45
00:02:19,520 --> 00:02:23,780
‫So why does distribution distributor transaction exist?

46
00:02:23,780 --> 00:02:24,170
‫Right.

47
00:02:24,410 --> 00:02:25,730
‫So what is this thing?

48
00:02:26,930 --> 00:02:32,750
‫Before we jump to distribute the transaction, let's briefly take some time to talk about transactions

49
00:02:32,750 --> 00:02:33,290
‫in general.

50
00:02:33,290 --> 00:02:37,070
‫I made a video about transactions and acid and all that stuff.

51
00:02:37,070 --> 00:02:38,210
‫I'm going to freeze it right here.

52
00:02:38,480 --> 00:02:39,230
‫Go check it out.

53
00:02:39,260 --> 00:02:41,690
‫I really recommend you understand these basics.

54
00:02:42,080 --> 00:02:46,520
‫I think every software engineer need to understand what acid is a tomasetti, consistency, isolation,

55
00:02:47,210 --> 00:02:47,810
‫durability.

56
00:02:47,810 --> 00:02:52,820
‫So that's why I made a full course, almost video one hour talking about these things.

57
00:02:52,820 --> 00:02:57,100
‫It's at a very critical topic to discuss as a versing.

58
00:02:57,110 --> 00:02:57,380
‫Right.

59
00:02:58,390 --> 00:02:59,950
‫A transaction is.

60
00:03:00,920 --> 00:03:06,560
‫And I'm going to take the example that I'm seeing all the time is this order payment kind of thing.

61
00:03:06,570 --> 00:03:11,250
‫Let's say you're you're building Amazon or building some application that does ordering and payment,

62
00:03:11,280 --> 00:03:11,550
‫OK?

63
00:03:12,020 --> 00:03:14,330
‫And it's a Web application.

64
00:03:14,330 --> 00:03:21,860
‫So we have one Web server or WAP framework, maybe a Norges application or a B application that does

65
00:03:21,860 --> 00:03:22,930
‫the whole thing for you.

66
00:03:22,940 --> 00:03:23,270
‫Right.

67
00:03:23,300 --> 00:03:32,960
‫You submit a request, arrest maybe or or graft you up and you say, hey, I am I want to I have these

68
00:03:32,960 --> 00:03:33,710
‫in my cart.

69
00:03:33,710 --> 00:03:36,920
‫I want to buy the stuff.

70
00:03:36,920 --> 00:03:37,280
‫Right.

71
00:03:37,730 --> 00:03:39,890
‫So what is by this stuff mean?

72
00:03:39,930 --> 00:03:49,610
‫It means that submit an order, charge my credit card and update the order and email me with the order

73
00:03:49,610 --> 00:03:53,540
‫details and when is going to be shipped and all that jazz.

74
00:03:54,560 --> 00:03:54,910
‫All right.

75
00:03:55,010 --> 00:03:56,040
‫How do we do it today?

76
00:03:56,420 --> 00:03:57,410
‫Very simple.

77
00:03:57,650 --> 00:03:58,490
‫Very elegant.

78
00:03:58,500 --> 00:04:00,980
‫We figured it out back in the 70s.

79
00:04:00,980 --> 00:04:02,870
‫How to do this transactions right.

80
00:04:03,380 --> 00:04:04,600
‫In an atomic manner.

81
00:04:04,730 --> 00:04:10,550
‫That means all of these updates have to happen at the same time or none of them happen otherwise.

82
00:04:10,550 --> 00:04:13,960
‫We lead into these inconsistent state, which aren't that OK?

83
00:04:16,170 --> 00:04:19,650
‫Cotton gone bad, not bad for overapplication, but bad in this case.

84
00:04:20,310 --> 00:04:21,930
‫All right, so how do we build it?

85
00:04:21,960 --> 00:04:22,380
‫Well.

86
00:04:23,400 --> 00:04:30,030
‫I am going to receive the request, this is something I want to buy these items, so I'm going to insert

87
00:04:30,030 --> 00:04:33,000
‫these items and I'm going to create an order.

88
00:04:33,330 --> 00:04:36,060
‫I'm going to create a bunch of items.

89
00:04:36,270 --> 00:04:36,630
‫Right.

90
00:04:36,630 --> 00:04:39,420
‫And insert these items and do the orders or there's a row.

91
00:04:39,420 --> 00:04:42,390
‫And there was another maybe another table or other items, whatever.

92
00:04:42,690 --> 00:04:44,520
‫And then item one item to item three.

93
00:04:44,790 --> 00:04:51,750
‫And I'm going to update the status of them of the order saying, oh, this is initialized now and then

94
00:04:51,750 --> 00:04:57,510
‫I'm going to go ahead and charge my the payment method.

95
00:04:58,320 --> 00:04:59,610
‫You're using MasterCard.

96
00:04:59,620 --> 00:05:01,050
‫OK, charge it.

97
00:05:01,320 --> 00:05:09,000
‫If I succeed, I am going to move in and maybe estimate the delivery, say, OK, it's going to take

98
00:05:09,000 --> 00:05:16,110
‫a week to deliver and then I'm going to update the order status to be paid because I another update

99
00:05:16,110 --> 00:05:23,730
‫statements or the first insert and insert, maybe insert the payment method because I submitted it,

100
00:05:24,810 --> 00:05:26,850
‫I just inserted it, not committed yet.

101
00:05:27,210 --> 00:05:34,200
‫And then I update the order and then finally because the payment, all of these statements are now true

102
00:05:34,200 --> 00:05:36,510
‫and correct, I'm going to commit.

103
00:05:36,900 --> 00:05:37,260
‫Right.

104
00:05:37,350 --> 00:05:43,420
‫And this is a whole bunch of atomic operations now.

105
00:05:43,440 --> 00:05:47,790
‫This is one atomic operation, one transaction, OK, work, one unit of work.

106
00:05:48,510 --> 00:05:51,330
‫Well, and that's what happened.

107
00:05:51,330 --> 00:05:53,250
‫If the payments failed, well, the payments failed.

108
00:05:53,250 --> 00:05:56,100
‫I want to roll back the whole thing, like the order didn't exist.

109
00:05:58,320 --> 00:06:01,090
‫And just emailed that he was the hey failed.

110
00:06:01,130 --> 00:06:04,800
‫I'm not going to even submit the order because your payment didn't charge a while.

111
00:06:04,980 --> 00:06:06,300
‫Why would I create an order?

112
00:06:06,420 --> 00:06:07,490
‫Just useless.

113
00:06:07,500 --> 00:06:07,680
‫Right?

114
00:06:08,820 --> 00:06:10,710
‫So that's that's a technical operation.

115
00:06:10,710 --> 00:06:10,980
‫Right.

116
00:06:11,250 --> 00:06:13,590
‫And you can do all of that stuff because all of this.

117
00:06:14,930 --> 00:06:20,270
‫Work or methods or this is the same process that is running, it's either a Web server running or maybe

118
00:06:20,390 --> 00:06:26,150
‫even a veteran, a client server back when he was a client server application, you said you have one

119
00:06:26,150 --> 00:06:28,670
‫process running and it's doing this thing.

120
00:06:28,880 --> 00:06:35,900
‫And the TCP connection between you and the database is one TCP connection dedicated for this transaction.

121
00:06:38,160 --> 00:06:39,760
‫Easy peasy, right?

122
00:06:40,260 --> 00:06:49,140
‫We out how to do that, meet micro services, micro services came in and says, no, these big applications

123
00:06:49,140 --> 00:06:54,330
‫that does too many things are bad because they don't not scale.

124
00:06:55,050 --> 00:07:03,570
‫Because we cannot we can't we don't have the luxury of picking up my own languages as team.

125
00:07:03,570 --> 00:07:06,540
‫All the teams are forced to use the same programming language.

126
00:07:06,840 --> 00:07:15,410
‫And these hip kids these days like to use JavaScript and Rust and what other languages these days go.

127
00:07:15,420 --> 00:07:17,790
‫Yeah, everybody used to live in new languages.

128
00:07:17,790 --> 00:07:18,010
‫Right?

129
00:07:18,030 --> 00:07:18,520
‫So no.

130
00:07:19,080 --> 00:07:21,300
‫So meet Microsoft says let's break things up.

131
00:07:22,080 --> 00:07:23,280
‫We have problems with that.

132
00:07:23,290 --> 00:07:25,380
‫Obviously, I discussed most of these problems.

133
00:07:26,650 --> 00:07:30,060
‫The side Microsoft vs. whites good, whites bad.

134
00:07:30,840 --> 00:07:34,350
‫But regardless, it's there it's here to stay microdevices.

135
00:07:34,350 --> 00:07:37,300
‫It's painful, but it gives you some advantages.

136
00:07:37,320 --> 00:07:42,420
‫So let's discuss what problems arise with Microsoft as will come back to this.

137
00:07:42,420 --> 00:07:47,610
‫My ordering courting Ramezani kind of like system.

138
00:07:47,850 --> 00:07:51,670
‫Well, the order code should live on its own server as well.

139
00:07:52,140 --> 00:07:56,730
‫That's been up and OJEA servers and the spin up by order service.

140
00:07:56,730 --> 00:07:58,710
‫And this is just for creating orders.

141
00:07:59,220 --> 00:08:02,520
‫And the payment service is a completely different team, right?

142
00:08:02,530 --> 00:08:02,720
‫Yeah.

143
00:08:02,730 --> 00:08:07,560
‫Maybe you outsource to, I don't know, square or stripe or something like that.

144
00:08:07,560 --> 00:08:07,830
‫Right.

145
00:08:07,830 --> 00:08:09,480
‫So that's another service itself.

146
00:08:09,720 --> 00:08:14,510
‫And then which is going to use for I you didn't see the benefits of this and then I don't know, the

147
00:08:15,420 --> 00:08:20,700
‫the shipment is another service and and you can see now these are now not one process.

148
00:08:20,700 --> 00:08:21,830
‫They are multiple processes.

149
00:08:21,830 --> 00:08:23,610
‫Stop the order service.

150
00:08:23,880 --> 00:08:26,850
‫Even if all of them talk to the same, that you still have the same problem.

151
00:08:27,060 --> 00:08:30,810
‫But most of them now with micro services have their own flavor of a database.

152
00:08:30,990 --> 00:08:40,800
‫So the order could use Postgres and and and the the payment could use Mongo DB and the shipment could

153
00:08:40,800 --> 00:08:41,340
‫use.

154
00:08:42,500 --> 00:08:44,190
‫OK, Sandra, who cares.

155
00:08:44,190 --> 00:08:44,570
‫Right.

156
00:08:44,580 --> 00:08:48,340
‫And good luck making one transaction between all these three databases.

157
00:08:48,360 --> 00:08:49,050
‫It's impossible.

158
00:08:49,230 --> 00:08:49,640
‫Right.

159
00:08:50,460 --> 00:08:55,680
‫So yeah, even if it's one database, it's impossible to make a transaction, a single transaction that

160
00:08:55,680 --> 00:08:57,470
‫you can roll back if something failed.

161
00:08:57,870 --> 00:08:59,880
‫So now let's run through our scenario.

162
00:09:00,180 --> 00:09:02,160
‫I supplement a beautiful order.

163
00:09:02,610 --> 00:09:03,870
‫I go to the order service.

164
00:09:03,870 --> 00:09:04,170
‫Right.

165
00:09:04,500 --> 00:09:09,660
‫Assuming there is some sort of an ingress or reverse proxy that receives that thing and funnel and fan

166
00:09:09,660 --> 00:09:12,130
‫out request accordingly.

167
00:09:12,190 --> 00:09:14,070
‫That depends how it's done.

168
00:09:14,070 --> 00:09:15,240
‫There are many ways to do it.

169
00:09:16,080 --> 00:09:23,670
‫But I go ahead and submit that order so that orders say, OK, order created succeeded.

170
00:09:24,180 --> 00:09:26,820
‫Go back now, submit the payment.

171
00:09:27,810 --> 00:09:29,220
‫I'm the reverse proxy by the way.

172
00:09:29,250 --> 00:09:30,330
‫Order succeed.

173
00:09:30,690 --> 00:09:31,830
‫Go do the payment.

174
00:09:31,830 --> 00:09:32,400
‫Succeed.

175
00:09:32,400 --> 00:09:32,790
‫Good.

176
00:09:33,360 --> 00:09:33,960
‫Come back.

177
00:09:34,440 --> 00:09:35,130
‫Do that.

178
00:09:35,670 --> 00:09:38,010
‫Go ahead and do the shipment succeed.

179
00:09:38,520 --> 00:09:39,390
‫Email the customer.

180
00:09:39,570 --> 00:09:42,330
‫So that's that's three transactions, right.

181
00:09:42,330 --> 00:09:49,250
‫It's not one transaction but logically in the vascular it is one transaction to the user who gives the

182
00:09:49,260 --> 00:09:51,690
‫user doesn't know what you are using and doesn't know.

183
00:09:51,810 --> 00:09:55,590
‫He doesn't know or she doesn't know that you're using a transactional microdevices back end.

184
00:09:55,800 --> 00:09:56,670
‫Who cares.

185
00:09:56,730 --> 00:09:57,120
‫Right.

186
00:09:57,780 --> 00:09:58,620
‫So you're doing that.

187
00:09:58,620 --> 00:10:00,540
‫And what happened is the payment now fail.

188
00:10:00,580 --> 00:10:02,880
‫All payments fail, rollback back.

189
00:10:03,360 --> 00:10:03,900
‫Tough luck.

190
00:10:03,900 --> 00:10:09,360
‫You can either roll back because that was a commit that order actually committed to the database.

191
00:10:09,690 --> 00:10:10,080
‫Right.

192
00:10:11,420 --> 00:10:16,640
‫And now you have to tell, oh, by the way, the payments failed, so, hey, order, can you just.

193
00:10:17,550 --> 00:10:19,380
‫Rollback does not rollback.

194
00:10:19,480 --> 00:10:26,730
‫So you have to do these ugly things that's compensating it, its it's like, oh, order, delete the

195
00:10:26,730 --> 00:10:33,240
‫order, actually do a physical delete to go ahead and delete the order, which is kind of expensive

196
00:10:33,240 --> 00:10:36,300
‫because now you have rows in the database that you're doing.

197
00:10:36,480 --> 00:10:38,430
‫You're making the database do multiple work.

198
00:10:38,940 --> 00:10:42,300
‫Obviously it's not so much, but still it's work.

199
00:10:43,120 --> 00:10:43,490
‫Right.

200
00:10:43,830 --> 00:10:47,970
‫And if your database actually having it's in a cemetery.

201
00:10:47,970 --> 00:10:54,270
‫So all of these actually whereas entries that in the in your database so you will see them in your SSD

202
00:10:54,270 --> 00:10:55,090
‫drive or wherever.

203
00:10:55,680 --> 00:10:58,190
‫So it's expensive nevertheless.

204
00:10:58,470 --> 00:10:58,790
‫Right.

205
00:10:59,850 --> 00:11:01,620
‫Go back and do all the same.

206
00:11:01,650 --> 00:11:03,280
‫So that's the problem.

207
00:11:03,660 --> 00:11:04,380
‫That's why.

208
00:11:05,390 --> 00:11:11,380
‫Distributed transactions was invented, my god, 11 minutes to just get to the point, right?

209
00:11:13,220 --> 00:11:14,680
‫That's why I cannot do this stuff.

210
00:11:14,710 --> 00:11:16,140
‫Individual stories, guys.

211
00:11:16,340 --> 00:11:17,210
‫It's very hard.

212
00:11:17,450 --> 00:11:22,970
‫And you guys told me that you want this content pulled from the YouTube story and actually made into

213
00:11:22,970 --> 00:11:23,300
‫a video.

214
00:11:23,340 --> 00:11:24,140
‫That's why I'm doing it.

215
00:11:24,860 --> 00:11:29,990
‫Me know if this is a good idea, if you like it or not, if the content is boring or you are trying

216
00:11:29,990 --> 00:11:31,820
‫to make it as entertaining as possible.

217
00:11:32,240 --> 00:11:37,460
‫Software engineering is not attractive for everybody, but that's why we have this transaction.

218
00:11:37,490 --> 00:11:39,610
‫Now let's jump into the solutions.

219
00:11:39,950 --> 00:11:43,460
‫What can we do to fix daing this thing?

220
00:11:44,030 --> 00:11:45,350
‫What what problems can we sell?

221
00:11:45,900 --> 00:11:48,800
‫Well, we talked about one of them.

222
00:11:49,080 --> 00:11:52,580
‫Let's talk about the best one, which is Google have invented.

223
00:11:52,850 --> 00:11:53,270
‫Right.

224
00:11:53,930 --> 00:11:56,060
‫Which is called atomic clocks.

225
00:11:56,660 --> 00:11:56,990
‫Right.

226
00:11:57,500 --> 00:12:03,320
‫Google decided that all of their database using they have a database called Spanner.

227
00:12:03,680 --> 00:12:10,050
‫And all of these databases, have they they they have the same exact time.

228
00:12:10,110 --> 00:12:12,800
‫OK, and if you make a transaction.

229
00:12:13,850 --> 00:12:20,270
‫They know exactly what order of these transaction that happened, so they know because the time is so

230
00:12:20,270 --> 00:12:25,970
‫accurate, they nailed the time to the nearly microsecond maybe I don't know the details about that.

231
00:12:26,990 --> 00:12:31,820
‫And they were using atomic clock to actually synchronize all the databases so that you can actually

232
00:12:31,820 --> 00:12:36,080
‫physically do a transaction and roll it back because you know the steps, right?

233
00:12:36,390 --> 00:12:37,170
‫So they figured it out.

234
00:12:37,190 --> 00:12:42,380
‫I don't know the details about that, but that's how they do distributed transactions there.

235
00:12:42,890 --> 00:12:48,110
‫Another method, obviously not everybody has Google and nobody for the atomic clocks.

236
00:12:48,410 --> 00:12:54,080
‫So what the other solution is, do this compensating edits.

237
00:12:54,080 --> 00:13:02,150
‫Right, which is we talked about it created order, do a payment of the famous failed, come back and

238
00:13:02,150 --> 00:13:03,560
‫go actually delete the order.

239
00:13:03,560 --> 00:13:07,670
‫So just for each operation that you do, reverse it, reverse it.

240
00:13:07,770 --> 00:13:10,340
‫You have to try to reverse a lot of work for you.

241
00:13:10,760 --> 00:13:17,330
‫Very complicated logic to write, but it could happen, but very prone to bugs, I think.

242
00:13:18,020 --> 00:13:20,510
‫So that's one way to do it the other way.

243
00:13:20,720 --> 00:13:27,380
‫Instead of just doing this, having one process, do this stuff for you, which is also prone to failure,

244
00:13:27,380 --> 00:13:33,950
‫because if this process dies, you lost your you you will be in an inconsistent state, which is the

245
00:13:33,950 --> 00:13:36,830
‫second one, which I like a lot, the eventing model.

246
00:13:37,130 --> 00:13:38,330
‫So the eventing model.

247
00:13:39,540 --> 00:13:48,890
‫Coupled up with the what the what what's called with the compensating edit model SILLIAC.

248
00:13:49,790 --> 00:13:51,100
‫Use COFCO, go in.

249
00:13:51,140 --> 00:14:00,020
‫Q And when you publish when you create an order, the order service actually writes to that.

250
00:14:00,020 --> 00:14:03,230
‫Q Or the public or a Kafka topic.

251
00:14:03,590 --> 00:14:05,320
‫And by the way, I talked about cough, cough.

252
00:14:05,330 --> 00:14:06,110
‫Check it out here.

253
00:14:07,050 --> 00:14:14,200
‫Intellichoice Topinka says order has been created and people, people, services listen to this topic

254
00:14:14,210 --> 00:14:18,740
‫and says, oh, look, created, it's my job as a payment service to pick up this order that I created

255
00:14:18,740 --> 00:14:19,550
‫and start to pay it.

256
00:14:19,730 --> 00:14:25,790
‫And if it paid it, what it does is says, OK, if it successful, it writes another topic, says payment

257
00:14:25,790 --> 00:14:27,530
‫order paid successfully.

258
00:14:27,860 --> 00:14:32,540
‫So the shipment service will listen to that and we'll start listening to that and pick it up the order

259
00:14:32,810 --> 00:14:34,580
‫and if the payments failed.

260
00:14:36,440 --> 00:14:42,110
‫It will write another topic, it will always write the topic and says, hey, payments failed and the

261
00:14:42,110 --> 00:14:46,940
‫older service actually listens to the payment and to actually delete that, I think this is more elegant,

262
00:14:46,940 --> 00:14:48,620
‫event driven architecture.

263
00:14:49,010 --> 00:14:50,200
‫A lot of people talk about it.

264
00:14:50,210 --> 00:14:55,850
‫I think it's a beautiful design, to be honest, and it's a really nice, elegant design to do this

265
00:14:55,850 --> 00:14:56,060
‫thing.

266
00:14:56,060 --> 00:15:00,050
‫And Kafka esque Kafka with distribution transaction people.

267
00:15:00,050 --> 00:15:06,200
‫A lot more people are moving towards this architecture a lot because it's very elegant and nice.

268
00:15:07,400 --> 00:15:10,400
‫Kind of has it has its own decoupling kind of a thing.

269
00:15:10,620 --> 00:15:10,830
‫Right.

270
00:15:11,150 --> 00:15:17,360
‫You still have the Kafka queue or the broker that does all that stuff.

271
00:15:17,360 --> 00:15:24,770
‫But Robert complicated your stuff because you have to manage it out of Kafka is the one database.

272
00:15:24,770 --> 00:15:28,700
‫But and the final solution is, believe it or not.

273
00:15:29,950 --> 00:15:38,350
‫Go back to the old days, merge those dang services that are supposed to talk to the same database and

274
00:15:38,350 --> 00:15:42,740
‫they are supposed to be homogeneous transactions in the same database, in the same service.

275
00:15:43,150 --> 00:15:47,540
‫So go back to a mini monolith or a macro service.

276
00:15:47,590 --> 00:15:49,420
‫It's a macro service in general.

277
00:15:49,420 --> 00:15:49,720
‫Right.

278
00:15:49,720 --> 00:15:52,760
‫Or a mini monolith, whatever you want to call it.

279
00:15:52,960 --> 00:15:57,490
‫So the order and the shipment and the payment becomes the one service again.

280
00:15:57,490 --> 00:15:57,880
‫Right.

281
00:15:58,360 --> 00:16:07,630
‫But still, it's not does include every other things like shopping and cart service and product search

282
00:16:07,630 --> 00:16:08,950
‫and doesn't include all of that stuff.

283
00:16:08,950 --> 00:16:09,180
‫Right.

284
00:16:09,370 --> 00:16:14,850
‫Includes all the stuff that everything kind of does the touching the database in a transactional manner.

285
00:16:14,860 --> 00:16:20,430
‫So they use all the three methods that people are using to ensure distributed a transaction.

286
00:16:20,440 --> 00:16:27,010
‫And when you say distributed transaction is is a transaction that spans multiple databases or spans

287
00:16:27,010 --> 00:16:34,600
‫multiple services, multiple processes, not just A.D.A processes, doesn't have to be multiple services

288
00:16:34,600 --> 00:16:37,120
‫on different computers or different containers.

289
00:16:37,300 --> 00:16:41,230
‫They could be multiple processes in the same machine.

290
00:16:41,890 --> 00:16:44,700
‫So that's kind of distributed in the main way.

291
00:16:45,460 --> 00:16:52,120
‫It's not really horizontally distributed, but well, you can still you cannot achieve multiple transactions

292
00:16:52,120 --> 00:16:54,010
‫with from different processes anyway.

293
00:16:54,010 --> 00:16:56,140
‫So that kind of fall on the same track.

294
00:16:56,770 --> 00:16:58,600
‫Let me know if that was useful.

295
00:17:00,310 --> 00:17:04,210
‫Let me know if I said anything wrong or correct me if I said anything.

296
00:17:04,810 --> 00:17:06,670
‫That doesn't really sound right.

297
00:17:07,030 --> 00:17:09,910
‫I'm not really sure about the Google atomic clock.

298
00:17:09,910 --> 00:17:12,310
‫I need to read more about it, to be honest.

299
00:17:13,240 --> 00:17:17,200
‫But that was a summary of the transaction.

300
00:17:18,500 --> 00:17:19,320
‫What do you think, guys?

301
00:17:19,850 --> 00:17:26,150
‫Would you rather use a monolith and simplify your life or.

302
00:17:27,160 --> 00:17:28,120
‫Would you rather.

303
00:17:29,250 --> 00:17:35,370
‫Sacrifice and go to micro transactions and figure out how to distribute the transaction and figure out

304
00:17:35,370 --> 00:17:41,400
‫how to minimize your latency, because now you spurt things down, you have network calls, you have

305
00:17:41,400 --> 00:17:46,620
‫to figure that out and you have to also figure out how to debug your applications because debugging

306
00:17:46,620 --> 00:17:47,460
‫is hard as well.

307
00:17:47,940 --> 00:17:48,290
‫Right?

308
00:17:48,480 --> 00:17:51,450
‫Because now the bug is no longer but a break.

309
00:17:51,450 --> 00:17:55,170
‫But it's not it's not the same thing.

310
00:17:55,170 --> 00:17:55,370
‫Right.

311
00:17:55,600 --> 00:18:02,790
‫It's you have to put a attach a break debugger to this service and you need to know which exactly which

312
00:18:02,790 --> 00:18:07,010
‫process to attach it to and also go to the other servers and attach the debugger.

313
00:18:07,140 --> 00:18:13,170
‫And God knows that you hopefully you have one service of each one instance of each server so you can

314
00:18:13,170 --> 00:18:14,380
‫nail debugging is hard.

315
00:18:15,000 --> 00:18:23,970
‫So Data Dog and other providers and vendors are coming up with solutions like tracing and all that stuff.

316
00:18:23,970 --> 00:18:28,950
‫Try to minimize, simplify these kind of things, but still micro services.

317
00:18:30,510 --> 00:18:34,870
‫Communities, you have to do Bernardes, you have to manage all these daing containers, right?

318
00:18:35,290 --> 00:18:40,720
‫So, yeah, yeah, I'm just very interested to see what where will micro services lead us.

319
00:18:41,190 --> 00:18:41,490
‫Right.

320
00:18:42,550 --> 00:18:43,720
‫So great architecture.

321
00:18:43,990 --> 00:18:45,390
‫It's good to talk about it.

322
00:18:45,580 --> 00:18:50,470
‫It's a really beautiful thing for nerds like us to nerd on.

323
00:18:50,740 --> 00:18:56,290
‫And I just talk about Microsoft in general, but I don't know if it's feasible for every single one

324
00:18:56,290 --> 00:18:58,120
‫of us or one to build.

325
00:18:58,120 --> 00:19:01,140
‫Like Jenny, you want to build a donut shop.

326
00:19:02,200 --> 00:19:07,720
‫Does she really need to spend up like Copernicus's Kluster to host her Statik website?

327
00:19:08,590 --> 00:19:09,250
‫Probably not.

328
00:19:10,900 --> 00:19:13,260
‫Yeah, I just like I'm just interested.

329
00:19:13,280 --> 00:19:16,900
‫But the Googles and the Twitterers of the ward and the Facebook of the world.

330
00:19:16,900 --> 00:19:17,290
‫Yeah.

331
00:19:17,590 --> 00:19:24,610
‫They probably need this kind of micro services distributed so they can scale to the billions right off

332
00:19:24,610 --> 00:19:25,260
‫request.

333
00:19:25,930 --> 00:19:30,220
‫Yeah, I'm just very interested to know like how much would a monolith.

334
00:19:30,760 --> 00:19:31,150
‫I don't.

335
00:19:31,150 --> 00:19:40,300
‫Twitter reached a stage where they had to break down their micro services, their monolith they call

336
00:19:40,300 --> 00:19:42,820
‫a monorail because it was built on rails.

337
00:19:43,540 --> 00:19:43,960
‫Right.

338
00:19:44,530 --> 00:19:46,420
‫Because they couldn't scale it out.

339
00:19:47,140 --> 00:19:54,490
‫They that monolith was so big, spinning it up was so slow and just slow.

340
00:19:55,510 --> 00:20:00,100
‫It doesn't it couldn't keep up with the requests, especially back in the World Cup.

341
00:20:00,100 --> 00:20:03,070
‫2010 was it was year 2010.

342
00:20:03,250 --> 00:20:09,490
‫I don't remember every time one of the Germany will score a goal or any any team will score a goal.

343
00:20:09,730 --> 00:20:11,800
‫People will tweet Hegel.

344
00:20:11,800 --> 00:20:19,690
‫And the moment do they do that, Twitter goes down immediately because this flux of just comes off all

345
00:20:19,690 --> 00:20:24,640
‫over the world and the same exact second millisecond.

346
00:20:25,600 --> 00:20:25,980
‫Right.

347
00:20:26,320 --> 00:20:31,810
‫So, OK, let's let's just spin a more pvm there's been a more, but they couldn't keep up spinning

348
00:20:31,810 --> 00:20:37,050
‫up more VIMS just spinning a single VM to hose this monolith just a couple of seconds.

349
00:20:37,060 --> 00:20:41,650
‫And by the time it's spin up, it already filled up with requests, so.

350
00:20:43,100 --> 00:20:49,580
‫They needed to break it down into micro services so they can scale the part that actually is busy and

351
00:20:49,580 --> 00:20:53,660
‫instead of scanning the whole application, that is not like I mean, nobody's updating their profiles

352
00:20:53,660 --> 00:20:54,440
‫in the World Cup.

353
00:20:54,440 --> 00:20:54,780
‫Right?

354
00:20:55,130 --> 00:20:55,500
‫Right.

355
00:20:55,880 --> 00:21:02,120
‫The profile service just remains like in the single instance, whereas the actual rights service, whatever

356
00:21:02,120 --> 00:21:06,650
‫it's called, the tweet service will will receive more instances.

357
00:21:06,650 --> 00:21:06,840
‫Right.

358
00:21:06,890 --> 00:21:08,780
‫So that's essentially where the how they did it.

359
00:21:08,780 --> 00:21:10,070
‫And then it is very smartly.

360
00:21:11,930 --> 00:21:13,790
‫All right, guys, that's it for me today.

361
00:21:13,820 --> 00:21:16,280
‫Hope you enjoy this video and I'm going to see you in the next one.

362
00:21:16,520 --> 00:21:19,320
‫What should I talk about next in the comments section below?

363
00:21:19,640 --> 00:21:21,950
‫Give it a like subscribe if you didn't.

364
00:21:22,280 --> 00:21:22,730
‫And.

365
00:21:23,950 --> 00:21:24,710
‫Soon, the next one.

366
00:21:25,270 --> 00:21:27,250
‫You guys stay safe, stay awesome.