1
00:00:00,400 --> 00:00:03,170
‫Now let's discuss AWS step functions.

2
00:00:03,170 --> 00:00:04,510
‫So step functions allow you

3
00:00:04,510 --> 00:00:06,970
‫to model a workflow as state machines

4
00:00:06,970 --> 00:00:10,760
‫and you have one state machine per workflow.

5
00:00:10,760 --> 00:00:12,020
‫So this could be used if you want

6
00:00:12,020 --> 00:00:14,770
‫to have an order fulfillment or data processing,

7
00:00:14,770 --> 00:00:17,310
‫a web application or any workflow you want.

8
00:00:17,310 --> 00:00:19,460
‫The idea is that you will define

9
00:00:19,460 --> 00:00:21,880
‫what happens and what happens next,

10
00:00:21,880 --> 00:00:23,590
‫and then what happens based on some condition

11
00:00:23,590 --> 00:00:25,950
‫and so on to define a workflow.

12
00:00:25,950 --> 00:00:27,520
‫So as you can see on the right hand side,

13
00:00:27,520 --> 00:00:28,880
‫there's a visual workflow.

14
00:00:28,880 --> 00:00:31,330
‫We start and we have an example,

15
00:00:31,330 --> 00:00:32,310
‫is it yes or no?

16
00:00:32,310 --> 00:00:33,890
‫If it's no, then you go to the end.

17
00:00:33,890 --> 00:00:35,940
‫If it's yes, then you wait three seconds.

18
00:00:35,940 --> 00:00:37,520
‫Then you say, "Hello" and "World,"

19
00:00:37,520 --> 00:00:39,140
‫you compile the result into "Hello World"

20
00:00:39,140 --> 00:00:40,370
‫and you go into end.

21
00:00:40,370 --> 00:00:42,990
‫So we'll be going through hands-on to make it specific

22
00:00:42,990 --> 00:00:46,250
‫but the idea is that you define a workflow in JSON

23
00:00:46,250 --> 00:00:47,970
‫and this JSON allows you to get

24
00:00:47,970 --> 00:00:49,700
‫a visualization of the workflow

25
00:00:49,700 --> 00:00:51,760
‫and then it gets executed visually,

26
00:00:51,760 --> 00:00:55,430
‫and then we can see the history of the execution itself.

27
00:00:55,430 --> 00:00:57,040
‫So the idea is that every little step

28
00:00:57,040 --> 00:00:58,950
‫in the workflow is going to be, for example,

29
00:00:58,950 --> 00:01:02,150
‫a Lambda function or inserting data into DynamoDB

30
00:01:02,150 --> 00:01:03,940
‫or a specific ECS task,

31
00:01:03,940 --> 00:01:05,440
‫and the step function is going to be

32
00:01:05,440 --> 00:01:08,680
‫the thing that orchestrates this workflow for you.

33
00:01:08,680 --> 00:01:12,240
‫To start a workflow, you can use an SDK API call,

34
00:01:12,240 --> 00:01:15,410
‫you can use API gateway, you can use CloudWatch Events

35
00:01:15,410 --> 00:01:18,220
‫or Amazon EventBridge, they're the same thing,

36
00:01:18,220 --> 00:01:19,830
‫or you can launch, obviously,

37
00:01:19,830 --> 00:01:21,690
‫a step function manually from the console,

38
00:01:21,690 --> 00:01:23,960
‫as we will be doing it in the hands-on.

39
00:01:23,960 --> 00:01:26,950
‫So step functions have a bunch of boxes

40
00:01:26,950 --> 00:01:28,360
‫as you see on the right hand side

41
00:01:28,360 --> 00:01:30,660
‫and these boxes are called tasks.

42
00:01:30,660 --> 00:01:33,530
‫And so the task state is used to do some work

43
00:01:33,530 --> 00:01:34,980
‫in your state machine.

44
00:01:34,980 --> 00:01:38,490
‫So the task state could be invoking a Lambda function

45
00:01:38,490 --> 00:01:42,060
‫or it could be invoking a Batch job

46
00:01:42,060 --> 00:01:45,340
‫or running an ECS task and wait for it to complete

47
00:01:45,340 --> 00:01:47,360
‫or to insert an item directly

48
00:01:47,360 --> 00:01:49,700
‫from step functions into the DynamoDB

49
00:01:49,700 --> 00:01:52,670
‫or to publish a message into SNS or SQS

50
00:01:52,670 --> 00:01:55,730
‫or even to launch another step function workflow.

51
00:01:55,730 --> 00:01:58,540
‫All these things can be defined in a task state,

52
00:01:58,540 --> 00:02:02,080
‫or a task could also be running one activity

53
00:02:02,080 --> 00:02:05,070
‫and that activity could be an EC2 machine

54
00:02:05,070 --> 00:02:07,650
‫or an Amazon ECS task or on-premises server

55
00:02:07,650 --> 00:02:09,380
‫that is going to be pulling for some work

56
00:02:09,380 --> 00:02:11,220
‫from step functions to do the work

57
00:02:11,220 --> 00:02:13,930
‫and then send back the results to the step function.

58
00:02:13,930 --> 00:02:17,400
‫The idea is that here, we are more free in terms of that,

59
00:02:17,400 --> 00:02:20,090
‫the app server is not invoked by the step function,

60
00:02:20,090 --> 00:02:21,600
‫the step function at the app server

61
00:02:21,600 --> 00:02:23,850
‫is actually pulling for step functions

62
00:02:23,850 --> 00:02:25,920
‫to find activities and work to do.

63
00:02:25,920 --> 00:02:27,760
‫Which is more similar to how

64
00:02:27,760 --> 00:02:31,960
‫a service called SWF worked in AWS.

65
00:02:31,960 --> 00:02:34,150
‫So I wanna make this super concrete.

66
00:02:34,150 --> 00:02:36,270
‫So let's say we want to define a task state

67
00:02:36,270 --> 00:02:38,260
‫to invoke a Lambda function.

68
00:02:38,260 --> 00:02:40,270
‫So we're going to define a JSON

69
00:02:40,270 --> 00:02:42,960
‫and as you can see, the name is "Invoke Lambda function,"

70
00:02:42,960 --> 00:02:47,190
‫the type is task and the resource is Lambda function invoke.

71
00:02:47,190 --> 00:02:48,900
‫Then the parameters is the function name,

72
00:02:48,900 --> 00:02:50,760
‫so we specify the function name,

73
00:02:50,760 --> 00:02:53,850
‫and we pass it on input payload to our Lambda function.

74
00:02:53,850 --> 00:02:57,420
‫Then we say next, to go to the next state, and the time out,

75
00:02:57,420 --> 00:03:00,260
‫to see if our Lambda function is timing out.

76
00:03:00,260 --> 00:03:03,800
‫So what can be states in step functions?

77
00:03:03,800 --> 00:03:05,430
‫We could have this Choice State,

78
00:03:05,430 --> 00:03:08,380
‫to test for a condition to send it to a branch

79
00:03:08,380 --> 00:03:09,600
‫or the default branch.

80
00:03:09,600 --> 00:03:11,480
‫It could be a Fail or Succeed State,

81
00:03:11,480 --> 00:03:14,000
‫to stop executing the step function workflow

82
00:03:14,000 --> 00:03:15,960
‫with a success or a failure.

83
00:03:15,960 --> 00:03:17,300
‫It could be a Pass State,

84
00:03:17,300 --> 00:03:20,510
‫simply to pass the input to the output

85
00:03:20,510 --> 00:03:23,650
‫or to inject some fixed data without performing some work.

86
00:03:23,650 --> 00:03:26,080
‫It could be a Wait State, to provide a delay

87
00:03:26,080 --> 00:03:27,130
‫for a certain amount of time

88
00:03:27,130 --> 00:03:30,110
‫or until a specified date and time.

89
00:03:30,110 --> 00:03:33,160
‫It could be a Map State, to dynamically iterate steps,

90
00:03:33,160 --> 00:03:37,190
‫or a Parallel State to begin parallel branches of execution.

91
00:03:37,190 --> 00:03:38,950
‫What you need to remember from the exam perspective

92
00:03:38,950 --> 00:03:41,630
‫is most likely going to be the parallel states.

93
00:03:41,630 --> 00:03:43,450
‫So, and the task state

94
00:03:43,450 --> 00:03:45,360
‫that I showed you from before, obviously.

95
00:03:45,360 --> 00:03:47,920
‫So I wanna show you what happens visually.

96
00:03:47,920 --> 00:03:51,530
‫So we have a execution workflow and this is a workflow.

97
00:03:51,530 --> 00:03:52,440
‫So we submit a job.

98
00:03:52,440 --> 00:03:53,490
‫We wait for X seconds.

99
00:03:53,490 --> 00:03:54,430
‫We get the job status.

100
00:03:54,430 --> 00:03:55,263
‫Is the job complete?

101
00:03:55,263 --> 00:03:58,750
‫Yes or no, if not wait again, and so there's a loop.

102
00:03:58,750 --> 00:04:00,450
‫And when the job is complete,

103
00:04:00,450 --> 00:04:02,300
‫either you get the final job status

104
00:04:02,300 --> 00:04:04,750
‫because you completed successfully

105
00:04:04,750 --> 00:04:06,560
‫or if the job failed,

106
00:04:06,560 --> 00:04:09,010
‫then you maybe go and deal with that failure,

107
00:04:09,010 --> 00:04:10,900
‫and all of this go to the end.

108
00:04:10,900 --> 00:04:13,530
‫So as the execution is started,

109
00:04:13,530 --> 00:04:15,100
‫we're going to start, submit a job.

110
00:04:15,100 --> 00:04:16,237
‫We're gonna wait for X seconds

111
00:04:16,237 --> 00:04:18,640
‫and the blue is whatever is in progress.

112
00:04:18,640 --> 00:04:20,380
‫Then it goes into several loops

113
00:04:20,380 --> 00:04:22,190
‫and then the job status is done,

114
00:04:22,190 --> 00:04:24,880
‫and then we get the final job status and we go to the end,

115
00:04:24,880 --> 00:04:28,660
‫and this gives us a visual aspect of our step functions.

116
00:04:28,660 --> 00:04:29,950
‫So the step function is really is here

117
00:04:29,950 --> 00:04:32,070
‫to orchestrate whatever work you need to do,

118
00:04:32,070 --> 00:04:34,720
‫to define how your state machine is going to look like.

119
00:04:34,720 --> 00:04:36,530
‫What your workflow is going to look like,

120
00:04:36,530 --> 00:04:38,550
‫and I know that right now, this may be,

121
00:04:38,550 --> 00:04:39,870
‫what the hell is this talking about?

122
00:04:39,870 --> 00:04:40,920
‫This is very confusing.

123
00:04:40,920 --> 00:04:44,430
‫I can do all of that in just one Lambda function to program,

124
00:04:44,430 --> 00:04:46,950
‫but I will show you in the next lecture,

125
00:04:46,950 --> 00:04:48,870
‫a hands-on on step functions.

126
00:04:48,870 --> 00:04:50,420
‫It should make a lot more sense,

127
00:04:50,420 --> 00:04:52,940
‫and then I will show you the whole power of state functions

128
00:04:52,940 --> 00:04:56,040
‫in the further lecture by talking about error handling.

129
00:04:56,040 --> 00:04:58,790
‫So I will see you in the next lecture for the hands-on.