1
00:00:00,480 --> 00:00:01,680
High and won't come back.

2
00:00:01,770 --> 00:00:06,720
So let's continue our scaling you know Ruby's practice questions.

3
00:00:06,850 --> 00:00:07,270
Adventure

4
00:00:09,930 --> 00:00:16,810
evil although Skilling launch configuration is created via the command line interface which easy to

5
00:00:16,810 --> 00:00:18,470
monitoring will be enabled.

6
00:00:20,000 --> 00:00:24,830
To the other scanning launch configuration is created via a command line interface.

7
00:00:24,850 --> 00:00:29,260
Not it obvious console which easy to monitoring will be enabled.

8
00:00:29,330 --> 00:00:35,600
A detailed basic C monitoring will not be any level of command line interface was used to create the

9
00:00:35,600 --> 00:00:41,950
launch configuration the interface does not support creating launch configurations using command line

10
00:00:41,950 --> 00:00:42,300
interface.

11
00:00:42,310 --> 00:00:45,370
So this one is definitely wrong and this one is definitely wrong.

12
00:00:45,690 --> 00:00:53,330
So another question is command line interface used to create launch configuration for the scaling.

13
00:00:53,380 --> 00:00:56,460
Is that going to be detailed or basic what's the default.

14
00:00:56,560 --> 00:01:00,840
Basically So we know if it was a console that would have been.

15
00:01:00,910 --> 00:01:06,430
Yes that would have been the basic or every five minutes in the command line interface.

16
00:01:06,430 --> 00:01:08,280
The details will be the default.

17
00:01:08,290 --> 00:01:10,460
So this one is wrong and this is the correct answer

18
00:01:13,450 --> 00:01:19,040
so is correct because the command line enable detailed monitoring.

19
00:01:19,260 --> 00:01:27,130
White House counsel enables basic monitoring by default be incorrect because the basic is the default

20
00:01:27,150 --> 00:01:34,810
for it console created launch configurations seems incorrect because monitoring will or can be enabled

21
00:01:35,050 --> 00:01:38,300
if command line interface can is used.

22
00:01:38,510 --> 00:01:46,620
Andy is incorrect because it does support that so it can allow you to configure the notion of relation

23
00:01:46,630 --> 00:01:49,930
from the command line and the correct answer is.

24
00:01:49,990 --> 00:01:58,570
A quick refresher when you create launch configurations for auto scaling groups using command line interface.

25
00:01:58,570 --> 00:02:05,160
You have the detailed monitoring enabled by default whereas if you do that from the console then basic

26
00:02:05,160 --> 00:02:09,140
manhunting are every five minutes is the default.

27
00:02:09,370 --> 00:02:15,440
Keep that in mind when deciding the oldest killing policy for argument's killing.

28
00:02:15,460 --> 00:02:17,200
Alarm frequency configuration.

29
00:02:17,200 --> 00:02:22,120
It needs to match that of two instances in the launch configuration.

30
00:02:22,160 --> 00:02:27,250
So now if you are given a question that says that the launch configuration was created by command line

31
00:02:27,250 --> 00:02:34,010
interface what would be AWOS recommended setting for the cloud watch alarm.

32
00:02:34,210 --> 00:02:42,650
Then immediately you know it is 60 seconds not 300 seconds not 10 minutes and not one hour.

33
00:02:44,210 --> 00:02:50,640
So because the detailed monitoring will be enabled for the launch configuration that is configured using

34
00:02:50,640 --> 00:02:51,530
the command line interface.

35
00:02:51,530 --> 00:02:57,350
So instead of telling you all that they will only tell you that the command line interface was used

36
00:02:57,350 --> 00:03:03,310
to create the launch configuration expecting that you know this is detailed monitoring and not the basic

37
00:03:03,410 --> 00:03:08,020
monitoring next question.

38
00:03:08,180 --> 00:03:13,020
You are the new architect and you can do it to incorporate that.

39
00:03:13,100 --> 00:03:22,670
You have been assigned a task to design a highly available so high available fault tolerant it has architecture

40
00:03:22,700 --> 00:03:28,370
for a web layer of a three tier application in an NWS region.

41
00:03:28,850 --> 00:03:38,240
OK so you want highly available fault tolerant web tier of a three tier application in an either region.

42
00:03:38,480 --> 00:03:42,660
It is required that the affiliate of an availability zone.

43
00:03:42,800 --> 00:03:46,910
So now we're talking about the full tolerance and the high availability.

44
00:03:46,910 --> 00:03:56,360
So if one of these is on feels would not affect the food high availability the full available of the

45
00:03:56,440 --> 00:04:04,010
application under normal load the web requires 12 easy to optimize instances.

46
00:04:06,670 --> 00:04:08,440
So this is very tricky now.

47
00:04:08,890 --> 00:04:16,160
It is required that a failure of believe is if you lose one availability is on the failure of that system

48
00:04:16,210 --> 00:04:20,760
would not affect the full availability of the application.

49
00:04:20,770 --> 00:04:26,810
What is the full weight of the addiction that means the high availability of the application as well.

50
00:04:26,830 --> 00:04:31,990
How would you think could be the solution keeping cost effectiveness in mind so you may have multiple

51
00:04:31,990 --> 00:04:34,590
solutions you need to look at cost as well.

52
00:04:34,660 --> 00:04:40,240
So you need to ensure that a figure of fun is on will not impact the high availability of the solution

53
00:04:40,740 --> 00:04:42,310
or that we have.

54
00:04:42,720 --> 00:04:48,040
And also the solution must be cost effective so if you have multiple options choose the one that will

55
00:04:48,040 --> 00:04:50,370
cost less.

56
00:04:50,380 --> 00:04:51,710
Lets look at the options.

57
00:04:51,840 --> 00:04:57,030
A single region architecture fine single region is fine.

58
00:04:58,600 --> 00:05:02,710
Use one or two Skilling group in two availabilities zones.

59
00:05:02,800 --> 00:05:08,640
So what are the Skilling group Multi of celebrities on or just getting group into aggregate into these

60
00:05:08,650 --> 00:05:16,870
loans with a minimum desired capacity of 24 compute easy to instances and use an elastic load balancer

61
00:05:16,990 --> 00:05:20,510
that whoever it is owns to load balance between these two instances too.

62
00:05:20,520 --> 00:05:22,790
Let's think about it now.

63
00:05:22,840 --> 00:05:28,190
Is it commanding to launch a too easy otos killing.

64
00:05:28,220 --> 00:05:36,700
So I have an auto scaling group that will have a design capacity of 24 is two instances computer optimized.

65
00:05:36,710 --> 00:05:43,190
So that means it will try always strive to launch 12 and 12 to make it even and balanced between the

66
00:05:43,190 --> 00:05:52,670
two it is on and define an elastic load balancer under the killing group that will look after load balancing

67
00:05:52,670 --> 00:05:58,000
between the 24 instances and they will be will be also defined into availability zones.

68
00:05:58,340 --> 00:06:06,020
Looks wonderful but what happens if you lose one of availabilities abilities on the other one has enough

69
00:06:06,020 --> 00:06:08,150
instances to run the application right.

70
00:06:08,150 --> 00:06:17,690
They said we have 12 normal load requires 12 two instances but would this have an impact on the full

71
00:06:17,910 --> 00:06:20,890
high availability or the full availability of the application.

72
00:06:21,080 --> 00:06:22,430
Yes.

73
00:06:22,500 --> 00:06:28,940
So here I would expect for this even if I lose one of these on the solution will still although that

74
00:06:29,000 --> 00:06:31,590
it will still be highly available means what.

75
00:06:31,590 --> 00:06:36,200
Means more than one available it is on will be left even if I lose one.

76
00:06:36,360 --> 00:06:41,630
So what's the minimum to give me that is 1 to 3 every bit is on.

77
00:06:41,990 --> 00:06:51,170
So it is not the right one despite the fact it would be a single region architecture use one or two

78
00:06:51,170 --> 00:06:56,600
scaling group in three availabilities on so that are just getting group in this case will extend across

79
00:06:56,680 --> 00:06:57,110
three.

80
00:06:57,110 --> 00:07:03,430
Not to have elevated zones with a minimum design capacity of 18 easy to optimized instances.

81
00:07:03,470 --> 00:07:09,740
So 18 if the capacity is available in the free what will happen six will be launched here six will be

82
00:07:09,740 --> 00:07:11,340
launched here six will be launched here.

83
00:07:11,390 --> 00:07:11,900
Right.

84
00:07:11,900 --> 00:07:17,310
So that is 18 use and ill be in the three of them it is on.

85
00:07:17,310 --> 00:07:25,400
So define your be to the focus group and have it look after the three availabilities in this case to

86
00:07:25,790 --> 00:07:27,800
load balance between the two instances.

87
00:07:27,800 --> 00:07:30,080
Now lets see if this one is correct or wrong.

88
00:07:30,120 --> 00:07:32,410
If I lose one I believe it is what happens.

89
00:07:32,430 --> 00:07:39,780
I would have six here in six years that is 12 and that will give me high availability because we have

90
00:07:39,810 --> 00:07:45,080
two availabilities zones left so the 6 plus 6 is 12 that is fine.

91
00:07:45,210 --> 00:07:51,450
Hi everybody is not impacted and it is more cost effective than number one.

92
00:07:51,450 --> 00:07:52,430
Why then.

93
00:07:52,520 --> 00:07:52,920
Why.

94
00:07:52,920 --> 00:07:56,080
Because this is only 18 were as this was 24.

95
00:07:56,280 --> 00:07:59,410
So these have three checkmarks.

96
00:07:59,520 --> 00:08:04,170
Its going to give me that well is two instances it will remain highly available and its going to be

97
00:08:04,170 --> 00:08:13,510
cost effective See use route 53 and launch your web application across for availability zones into different

98
00:08:13,510 --> 00:08:19,000
regions and use elastic load balancer and auto Schelling group in each region with the desired minimum

99
00:08:19,000 --> 00:08:23,250
capacity of 8 easy to instances in each region.

100
00:08:23,260 --> 00:08:29,320
So here this is saying that you have two regions and you will have two availability zones in each region

101
00:08:30,220 --> 00:08:32,070
and use the route 53 as well.

102
00:08:32,110 --> 00:08:38,380
So route 53 will be outside and redrafted 3 will look at the Albee's and they always have the oldest

103
00:08:38,380 --> 00:08:40,240
getting groups in each region.

104
00:08:40,240 --> 00:08:42,890
So I have an ill be here with auto scaling.

105
00:08:42,920 --> 00:08:44,620
Ill be here with all the skin.

106
00:08:44,650 --> 00:08:52,140
Now this let's see if I lose one of these and what happens it is the two instances in each other's getting

107
00:08:52,140 --> 00:08:57,220
groups so this will be eight and this will be a total of 16 if I was one of them it is on I will have

108
00:08:57,220 --> 00:08:58,340
four remaining.

109
00:08:58,540 --> 00:09:04,150
No actually not four because if I lose one I get is on the other screening will kick in and try to launch

110
00:09:04,150 --> 00:09:07,660
it all the 8 here if I have eight design capacity.

111
00:09:07,780 --> 00:09:17,070
So I have an end to that 16 more than 12 but so this is 16 easy two instances and two others killing

112
00:09:17,110 --> 00:09:17,690
groups.

113
00:09:17,760 --> 00:09:20,890
Other groups dont cost money to help because they cost money.

114
00:09:20,900 --> 00:09:22,260
Route 53 costs money.

115
00:09:22,300 --> 00:09:24,290
So this is not cost effective.

116
00:09:24,610 --> 00:09:26,640
Thats why I will rule it out.

117
00:09:26,850 --> 00:09:30,790
You can't guarantee full availability in case of an easy failure and this is wrong.

118
00:09:30,790 --> 00:09:31,250
You can't.

119
00:09:31,270 --> 00:09:39,550
We just proved by B is that you can so the one that takes all the check all the all the boxes is B and

120
00:09:39,550 --> 00:09:41,940
that's why this is the correct answer.

121
00:09:45,770 --> 00:09:47,010
So is wrong.

122
00:09:47,030 --> 00:09:47,510
Why.

123
00:09:47,510 --> 00:09:52,870
Because it will work will not guarantee full age ain't an easy failure scenario only when everything

124
00:09:52,880 --> 00:09:53,960
is on will be up.

125
00:09:53,960 --> 00:10:00,270
If I lose one of the two B is correct because if one is the fails the other two would already have 12

126
00:10:00,280 --> 00:10:01,980
easy two instances up and running.

127
00:10:02,090 --> 00:10:04,570
And I believe it will be there as well.

128
00:10:04,790 --> 00:10:14,600
C is wrong because it's more costly plus route 53 cost and to Elby cost and took 16 easy to instances

129
00:10:14,630 --> 00:10:22,390
under normal operation and not a not 12 D is incorrect because you can guarantee full availability.

130
00:10:22,580 --> 00:10:30,230
So the correct answer is exactly b the fresher option a would have met the requirement if it was not

131
00:10:30,230 --> 00:10:37,190
for the maintain full high availability in the event of everybody's failure which indicates that the

132
00:10:37,190 --> 00:10:40,040
architecture must be over a minimum of two every bit as.

133
00:10:40,070 --> 00:10:42,470
Even in case of an easy failure.

134
00:10:42,470 --> 00:10:47,400
So that implies the availability zone's solution.

135
00:10:47,490 --> 00:10:54,390
Also the two entities on solution needs 24 is two instances as opposed to 18 in theory everybody is

136
00:10:54,390 --> 00:10:55,880
on solutions.

137
00:10:55,890 --> 00:11:00,960
This is why the three of them it is an option is preferred over the to everybody's own solution.

138
00:11:00,960 --> 00:11:06,210
That's why we have Big B and not a next question.

139
00:11:06,700 --> 00:11:08,680
You are the new architect.

140
00:11:08,710 --> 00:11:09,730
You can do it too.

141
00:11:09,730 --> 00:11:12,880
This is a tweak or at worst off the previous question.

142
00:11:12,880 --> 00:11:17,750
So I think it's going to be faster than the other one you have been assigned the task to design high

143
00:11:17,780 --> 00:11:19,740
available highly available faltering.

144
00:11:19,740 --> 00:11:24,300
It is architecture for a web layer of a three tier web application.

145
00:11:24,340 --> 00:11:31,810
In any ws region OK it is required that the failure of a very bit is on would not affect the full high

146
00:11:31,810 --> 00:11:33,960
availability of the web application.

147
00:11:34,050 --> 00:11:36,430
Two comments here before I continue here.

148
00:11:36,430 --> 00:11:41,390
They said would not affect what if the question was would affect.

149
00:11:41,410 --> 00:11:47,560
So that means to every bit is on is fine under normal load that requires 12 easy to instances.

150
00:11:47,560 --> 00:11:53,220
How would you Arctic the solution keeping cost effective itness in mind using single availabilities

151
00:11:53,230 --> 00:12:00,040
on auto scaling groups so here they are telling you the other groups should not span across it on each

152
00:12:00,040 --> 00:12:08,220
of his own will have its own skill killing a single region architecture used to single every bit is

153
00:12:08,220 --> 00:12:14,920
on auto scaling grouped into ISS with a minimum reserve capacity of 12 compute to instances in each.

154
00:12:14,950 --> 00:12:22,720
So the total is 24 hour news and it'll be in the 2 availability zones to load balance between the two

155
00:12:22,720 --> 00:12:24,100
instances.

156
00:12:24,100 --> 00:12:30,550
So this is exactly similar like in the previous scenario or the previous previous question.

157
00:12:30,550 --> 00:12:35,730
Except for the fact that we are using a single availabilities on auto scaling groups and not one or

158
00:12:35,740 --> 00:12:43,150
just getting group mapping across what differences give me I have here desired 12 years or 12 but if

159
00:12:43,150 --> 00:12:49,570
I lose one I lose the high availability so that's why I WILL NOT GO WITH be a single Reijer architecture

160
00:12:49,570 --> 00:12:55,420
use every single Asie all these killing groups in three of these zones with a minimum design capacity

161
00:12:55,420 --> 00:13:02,680
of 6 in each use an elastic lock balancer in three of elevated zones to load balance between the two

162
00:13:02,680 --> 00:13:04,490
instances in three phases.

163
00:13:04,540 --> 00:13:08,860
So that's exactly like the previous B in the previous question.

164
00:13:08,920 --> 00:13:11,130
I have 6 6 and 6.

165
00:13:11,140 --> 00:13:15,380
The difference is the artist and group were not spent across the three.

166
00:13:15,490 --> 00:13:19,930
I will have an audio skinning group for the first one or this killing you for the second one of getting

167
00:13:19,930 --> 00:13:26,980
them for the third one and then I haven't E-L be defined under these or just getting groups and they'll

168
00:13:27,000 --> 00:13:34,080
be able to look after the instances in each of their energies on losing one.

169
00:13:34,090 --> 00:13:37,480
I still have high availability and I have 12 which I need to be.

170
00:13:37,480 --> 00:13:41,030
Is a correct answer unless if I have a better answer.

171
00:13:41,110 --> 00:13:46,150
See yuzuru 53 and launch your web application across for Easy's into different regions.

172
00:13:46,150 --> 00:13:50,960
We know that this one is more costly and you can not guarantee yes you can so be.

173
00:13:50,980 --> 00:13:52,320
Again is the correct answer.

174
00:13:52,350 --> 00:13:58,850
But here I'm just showing you that if they play with a single easy Monte easy or just getting grooves

175
00:13:58,860 --> 00:14:02,920
should not cause you to freak out and find out OK what do you mean by this.

176
00:14:02,920 --> 00:14:03,860
Just think about it.

177
00:14:03,920 --> 00:14:04,620
Killing killing.

178
00:14:04,660 --> 00:14:07,880
If it is in one of its own it will function when everybody is on.

179
00:14:08,050 --> 00:14:10,160
If it is configured across is.

180
00:14:10,180 --> 00:14:12,720
It will do its magic across the previous ones.

181
00:14:12,730 --> 00:14:14,730
That's how easy it is.

182
00:14:14,770 --> 00:14:15,900
Same thing Puriel be.

183
00:14:15,910 --> 00:14:22,060
If it is defined in one if one if it isn't into the tracing you will be you have only one subnet defined

184
00:14:22,060 --> 00:14:28,450
for the iel be one public subnet in and everybody's own will be will only load balance to two instances

185
00:14:28,450 --> 00:14:30,080
in that it is.

186
00:14:30,550 --> 00:14:32,860
You wanted to function in another area it is on.

187
00:14:32,950 --> 00:14:38,260
Define another public subnet to that you be in that debate is on.

188
00:14:38,320 --> 00:14:39,680
That's how it works.

189
00:14:39,730 --> 00:14:46,120
So multi is the single azy should not freak you out as long as you understand that this is a region

190
00:14:46,120 --> 00:14:48,930
confined service but it can be across.

191
00:14:48,970 --> 00:14:54,040
It is not automatic to not deliberately stretch to cover other these zones.

192
00:14:54,060 --> 00:14:58,960
That's a manual thing you need to do unless you are able to mention tools but still someone would have

193
00:14:59,390 --> 00:15:03,990
clicked to stretch that or just getting group between Or just decide on its own.

194
00:15:04,090 --> 00:15:06,600
So again B is the correct answer here.

195
00:15:06,730 --> 00:15:13,090
The fact that you have now the full range even in case of an easy failure in this question and single

196
00:15:13,090 --> 00:15:19,780
event zone auto scaling groups the added number of total easy two instances.

197
00:15:19,840 --> 00:15:22,560
Two hundred and fifty or forty you need during normal load.

198
00:15:22,600 --> 00:15:27,150
That's why when we went through 3 is easy and we set the desired in 86 6.

199
00:15:27,160 --> 00:15:30,220
This is actually 150 percent of what we need.

200
00:15:30,220 --> 00:15:32,080
What we need is actually 12.

201
00:15:32,140 --> 00:15:33,660
Here I have 50 percent more.

202
00:15:33,670 --> 00:15:39,310
This means if one of these on fails you retain 100 percent of what is needed and that's why we said

203
00:15:39,690 --> 00:15:45,550
or the impact it should not impact your availability or high availability that means the same number

204
00:15:45,550 --> 00:15:48,130
of required is two instances should be there.

205
00:15:48,190 --> 00:15:56,360
What happens if you lost 2 to 80 zones and you have only 6 so normal operation requires 12.

206
00:15:56,350 --> 00:15:57,430
Now you have only 6.

207
00:15:57,460 --> 00:15:59,380
Does that impact your availability.

208
00:15:59,380 --> 00:16:00,580
Yes it does.

209
00:16:00,580 --> 00:16:02,320
Does it impact your response time.

210
00:16:02,320 --> 00:16:09,750
Yes it does does impact how your auto scaling would function because the desired desired is 6.

211
00:16:09,790 --> 00:16:14,930
It will not get any more launch any more is two instances.

212
00:16:15,040 --> 00:16:22,420
Except if there was an on demand scaling policy configured that says for instance when they reach 75

213
00:16:22,420 --> 00:16:25,290
percent you have to knock out.

214
00:16:25,450 --> 00:16:28,050
You have to launch into two instances.

215
00:16:28,090 --> 00:16:32,840
All right so let's take a break now and we'll come back to more auto scaling fun.

216
00:16:33,130 --> 00:16:33,900
I'll see you then.