1
00:00:07,265 --> 00:00:09,772
- First let's talk about
synchronizing development.

2
00:00:09,772 --> 00:00:11,454
Well what's an Agile Release Train?

3
00:00:11,454 --> 00:00:13,221
Well it's a bit of an abstract concept.

4
00:00:13,221 --> 00:00:15,558
It's illustrated here for sure.

5
00:00:15,558 --> 00:00:17,177
But it's really all of the above,

6
00:00:17,177 --> 00:00:21,606
so all of this stuff constitutes
the Agile Release Train.

7
00:00:21,606 --> 00:00:25,998
Teams, many teams, team
practices, team iterations,

8
00:00:25,998 --> 00:00:28,960
actual program implements
with PI planning,

9
00:00:28,960 --> 00:00:31,629
program execution at these boundaries

10
00:00:31,629 --> 00:00:33,763
using features of neighbors and then

11
00:00:33,763 --> 00:00:35,577
there processes continues exploration,

12
00:00:35,577 --> 00:00:37,557
continues integration,
continues appointment,

13
00:00:37,557 --> 00:00:38,627
that's what it all is.

14
00:00:38,627 --> 00:00:41,050
So it's really an
aggregation of all the teams

15
00:00:41,050 --> 00:00:43,843
and the key stake-holders
that need to deliver value.

16
00:00:43,843 --> 00:00:47,867
Long-lived, self-organizing
team of agile teams.

17
00:00:47,867 --> 00:00:49,888
So first for organization often now

18
00:00:49,888 --> 00:00:51,364
could be physical organization meaning

19
00:00:51,364 --> 00:00:53,128
it could be a line management organization

20
00:00:53,128 --> 00:00:54,233
or product organization.

21
00:00:54,233 --> 00:00:56,171
And for those that are of traditional

22
00:00:56,171 --> 00:00:58,225
kind of IT background
we start to think about

23
00:00:58,225 --> 00:00:59,988
the solutions as products,

24
00:00:59,988 --> 00:01:01,547
those typically move that way.

25
00:01:01,547 --> 00:01:03,474
But we call it virtual for reason because

26
00:01:03,474 --> 00:01:06,426
we're not sure exactly how
long-lived it's gonna be,

27
00:01:06,426 --> 00:01:09,750
we're creating a new construct
that's gonna deliver value,

28
00:01:09,750 --> 00:01:10,981
we're gonna discover that it can't

29
00:01:10,981 --> 00:01:12,675
do some of the things that it needs to do.

30
00:01:12,675 --> 00:01:14,844
50 to 125 people, why?

31
00:01:14,844 --> 00:01:18,741
Because it's fundamental kind
of law of social physics.

32
00:01:18,741 --> 00:01:21,121
That many people I can
know and care about,

33
00:01:21,121 --> 00:01:23,829
and below that number I can
sure know and care about them

34
00:01:23,829 --> 00:01:25,672
but I can't really build a big system.

35
00:01:25,672 --> 00:01:28,586
Above that number I can't
really know and care about them

36
00:01:28,586 --> 00:01:30,280
so I can't really build a big system.

37
00:01:30,280 --> 00:01:32,423
We're gonna have multiple ARTs to do that.

38
00:01:32,423 --> 00:01:35,835
We're gonna use this term called
a program increment, a PI.

39
00:01:35,835 --> 00:01:38,228
It's just a label for a fixed time box.

40
00:01:38,228 --> 00:01:40,156
Default in safe is 10 weeks,

41
00:01:40,156 --> 00:01:42,336
but that default is customizable.

42
00:01:42,336 --> 00:01:46,977
I've seen PIs as short as six,
maybe even four in one case,

43
00:01:46,977 --> 00:01:51,046
as long as 12, longer than
12, generally too long.

44
00:01:51,046 --> 00:01:52,976
To much change happens, right?

45
00:01:52,976 --> 00:01:55,246
Not enough opportunities
to synchronize again

46
00:01:55,246 --> 00:01:56,850
into a current plan of record.

47
00:01:56,850 --> 00:01:58,993
Within that the iterations
are synchronized

48
00:01:58,993 --> 00:02:02,438
and across trains the PIs
are synchronized true.

49
00:02:02,438 --> 00:02:04,084
They're aligned to a single mission

50
00:02:04,084 --> 00:02:06,425
via a single program
backlog, one program backlog.

51
00:02:06,425 --> 00:02:08,028
The team has their backlog for sure,

52
00:02:08,028 --> 00:02:09,834
all the local stuff they care about,

53
00:02:09,834 --> 00:02:11,638
but new stories in that backlog

54
00:02:11,638 --> 00:02:13,860
are driven by the program backlog.

55
00:02:13,860 --> 00:02:15,630
Architectural and UX guidance, how come?

56
00:02:15,630 --> 00:02:18,467
You really want from feature to feature

57
00:02:18,467 --> 00:02:21,223
the user to have an entirely
different user experience?

58
00:02:21,223 --> 00:02:22,338
I don't think so.

59
00:02:22,338 --> 00:02:25,341
Do we want this common single sign on

60
00:02:25,341 --> 00:02:29,423
across all the applications
in that tool suite?

61
00:02:29,423 --> 00:02:31,349
We absolutely do, so we're gonna provide

62
00:02:31,349 --> 00:02:33,441
architectural and UX guidance to help

63
00:02:33,441 --> 00:02:35,986
align the teams to a common outcome,

64
00:02:35,986 --> 00:02:37,304
a common way of working,

65
00:02:37,304 --> 00:02:38,706
common way of working for them

66
00:02:38,706 --> 00:02:40,900
and a common way of
working for their customer.

67
00:02:40,900 --> 00:02:43,582
And each Agile Release
Train frequently produces

68
00:02:43,582 --> 00:02:46,046
valuable and evaluable
system-level solutions.

69
00:02:46,046 --> 00:02:48,636
In other words they go through the steps

70
00:02:48,636 --> 00:02:50,191
of defining a functionality,

71
00:02:50,191 --> 00:02:51,468
implementing that functionality,

72
00:02:51,468 --> 00:02:53,162
acceptance testing and deploying that

73
00:02:53,162 --> 00:02:54,977
as quickly as they can.

74
00:02:54,977 --> 00:02:56,541
That's the purpose of an ART.

75
00:02:56,541 --> 00:02:58,478
We're gonna see how devops feeds into that

76
00:02:58,478 --> 00:03:01,064
because that's gonna be a consistent flow

77
00:03:01,064 --> 00:03:03,247
of small batches that
go through the system

78
00:03:03,247 --> 00:03:05,164
all the way to release.

79
00:03:06,590 --> 00:03:09,758
Now we talk about cadence
and synchronization,

80
00:03:09,758 --> 00:03:12,262
we need to have a little
further discussion here,

81
00:03:12,262 --> 00:03:14,364
which is that we need both.

82
00:03:14,364 --> 00:03:17,075
We need the teams to have the same cadence

83
00:03:17,075 --> 00:03:20,520
and the same synchronization
point, and here's why.

84
00:03:20,520 --> 00:03:23,351
If we don't have that,
and it's not unusual

85
00:03:23,351 --> 00:03:25,691
to have teams that maybe have

86
00:03:25,691 --> 00:03:27,089
a three week sprint length,

87
00:03:27,089 --> 00:03:28,533
or maybe this is a three
week sprint length,

88
00:03:28,533 --> 00:03:29,635
and this is two week sprint length

89
00:03:29,635 --> 00:03:31,321
and maybe this is a two week sprint length

90
00:03:31,321 --> 00:03:32,563
and that looks pretty good.

91
00:03:32,563 --> 00:03:34,501
So why should I bother to have them

92
00:03:34,501 --> 00:03:36,347
start and stop on the same date?

93
00:03:36,347 --> 00:03:40,178
Well, they're gonna develop
their work here and here,

94
00:03:40,178 --> 00:03:42,227
and we're gonna come to a point where

95
00:03:42,227 --> 00:03:43,664
we'd like to integrate.

96
00:03:43,664 --> 00:03:46,253
Well if we've done any
branches, even small branches,

97
00:03:46,253 --> 00:03:49,168
or even the integration requires work

98
00:03:49,168 --> 00:03:50,444
and it doesn't get done,

99
00:03:50,444 --> 00:03:51,638
for example maybe this integration

100
00:03:51,638 --> 00:03:53,536
requires work that
doesn't happen until here,

101
00:03:53,536 --> 00:03:55,635
it's impossible to synchronize that.

102
00:03:55,635 --> 00:03:59,223
So what happens is we have
this common anti-pattern,

103
00:03:59,223 --> 00:04:01,480
which is the teams are sprinting

104
00:04:01,480 --> 00:04:04,680
and they might even feel
good about that process,

105
00:04:04,680 --> 00:04:07,440
but the system isn't, the
system isn't sprinting.

106
00:04:07,440 --> 00:04:10,207
It's still in a monolithic,
waterfall-like world,

107
00:04:10,207 --> 00:04:11,969
it's not been integrated.

108
00:04:11,969 --> 00:04:14,189
So until we align those splints

109
00:04:14,189 --> 00:04:16,409
and get the ability to create kind of

110
00:04:16,409 --> 00:04:18,138
necessary force integration cycles,

111
00:04:18,138 --> 00:04:20,072
we just go with that the slowest component

112
00:04:20,072 --> 00:04:21,886
drags a train, which is always true,

113
00:04:21,886 --> 00:04:23,203
but most importantly that we

114
00:04:23,203 --> 00:04:25,013
don't have any integration points.

115
00:04:25,013 --> 00:04:27,698
The teams are sprinting, but
the system isn't sprinting.

116
00:04:27,698 --> 00:04:29,673
Common anti-pattern and a real issue

117
00:04:29,673 --> 00:04:33,161
and I'll bet anything you've
seen that in your organization.

118
00:04:33,161 --> 00:04:34,723
So instead we synchronize that,

119
00:04:34,723 --> 00:04:36,001
we synchronize those sprints,

120
00:04:36,001 --> 00:04:38,423
we make sure that the teams are lying

121
00:04:38,423 --> 00:04:40,068
on a common sprint life.

122
00:04:40,068 --> 00:04:41,581
We provide continuous integration

123
00:04:41,581 --> 00:04:42,973
because we're gonna need that

124
00:04:42,973 --> 00:04:44,568
to support a dev operated process.

125
00:04:44,568 --> 00:04:47,471
And now we probably also discover

126
00:04:47,471 --> 00:04:50,468
that it's very difficult
for a feature team

127
00:04:50,468 --> 00:04:53,430
to recessarily integrate the whole system.

128
00:04:53,430 --> 00:04:58,413
That might be true in a
small desktop application,

129
00:04:58,413 --> 00:05:01,411
it's not gonna be true in a
really, really large system.

130
00:05:01,411 --> 00:05:02,892
And the bigger the system is

131
00:05:02,892 --> 00:05:04,163
the more likely it is that we're

132
00:05:04,163 --> 00:05:05,360
gonna need help from them.

133
00:05:05,360 --> 00:05:08,241
And we have them helping a
little bit, running the gears,

134
00:05:08,241 --> 00:05:10,626
we can start making sure
that the system is sprinting

135
00:05:10,626 --> 00:05:12,303
and because it's a line we can

136
00:05:12,303 --> 00:05:14,848
add system demos at that point.

137
00:05:14,848 --> 00:05:16,497
And we'll talk a little later on

138
00:05:16,497 --> 00:05:18,754
about how what's essential
in essential safe

139
00:05:18,754 --> 00:05:21,470
and one of those things is
the presence of a system demo,

140
00:05:21,470 --> 00:05:23,154
so this gives us the tools we need now.

141
00:05:23,154 --> 00:05:24,960
So with a little help from the system team

142
00:05:24,960 --> 00:05:26,398
or maybe it's a group of individuals

143
00:05:26,398 --> 00:05:28,211
that came together
through the integration,

144
00:05:28,211 --> 00:05:29,818
maybe it's a different team entirely

145
00:05:29,818 --> 00:05:31,259
but that was on the train.

146
00:05:31,259 --> 00:05:33,686
And we can integrate and
integrate continuously

147
00:05:33,686 --> 00:05:35,666
and provide really fast feedback

148
00:05:35,666 --> 00:05:38,500
of the system, not just
the individual component

149
00:05:38,500 --> 00:05:41,081
or feature that the team is building.