1
00:00:00,200 --> 00:00:01,270
So, it's just a quick lecture,

2
00:00:01,270 --> 00:00:02,103
just to touch base

3
00:00:02,103 --> 00:00:04,290
on what is scalability and high availability.

4
00:00:04,290 --> 00:00:05,350
This is quite the beginners level,

5
00:00:05,350 --> 00:00:07,120
so if you feel very confident about these concepts,

6
00:00:07,120 --> 00:00:09,450
feel free to skip this lecture.

7
00:00:09,450 --> 00:00:11,550
Now, scalability means that your application system

8
00:00:11,550 --> 00:00:14,650
can hinder a greater load by adapting.

9
00:00:14,650 --> 00:00:16,670
And so, there are two kind of scalability.

10
00:00:16,670 --> 00:00:18,740
There's going to be vertical scalability

11
00:00:18,740 --> 00:00:22,650
or horizontal scalability, also called elasticity.

12
00:00:22,650 --> 00:00:26,160
And so, scalability is different from high availability.

13
00:00:26,160 --> 00:00:28,520
They're linked but different.

14
00:00:28,520 --> 00:00:31,340
So, what I want to do is deep dive into all this distinction

15
00:00:31,340 --> 00:00:33,660
and we'll use a call center as a fun example

16
00:00:33,660 --> 00:00:36,030
to really put into practice how things work.

17
00:00:36,030 --> 00:00:38,470
So, let's talk about vertical scalability.

18
00:00:38,470 --> 00:00:39,400
Vertical scalability,

19
00:00:39,400 --> 00:00:40,520
that means that you need

20
00:00:40,520 --> 00:00:43,180
to increase the size of your instance.

21
00:00:43,180 --> 00:00:45,500
So, let's take a phone operator, for example.

22
00:00:45,500 --> 00:00:47,620
We have a junior operator and we just hired him.

23
00:00:47,620 --> 00:00:51,070
He's great but he can only take five calls per minute.

24
00:00:51,070 --> 00:00:54,730
Now, we have a senior operator and he's much greater.

25
00:00:54,730 --> 00:00:57,720
He can take up to ten calls per minute.

26
00:00:57,720 --> 00:01:00,760
So, we've basically scaled up our junior operator

27
00:01:00,760 --> 00:01:03,830
into senior operator and he's faster and better.

28
00:01:03,830 --> 00:01:05,290
This is vertical scalability.

29
00:01:05,290 --> 00:01:07,630
As you can see, it goes up.

30
00:01:07,630 --> 00:01:09,260
So, for example, in EC2,

31
00:01:09,260 --> 00:01:11,360
our application runs on a a t2.micro

32
00:01:11,360 --> 00:01:13,210
and we want to upscale that application,

33
00:01:13,210 --> 00:01:16,000
that means we want to run it on a t2.large

34
00:01:16,000 --> 00:01:19,090
So, when do we use vertical scalability?

35
00:01:19,090 --> 00:01:22,120
Well, it's very common when you have non distributed system,

36
00:01:22,120 --> 00:01:23,510
such as a data base.

37
00:01:23,510 --> 00:01:25,650
So, it's quite common for a data base, for example,

38
00:01:25,650 --> 00:01:27,300
on RDS or ElastiCache,

39
00:01:27,300 --> 00:01:30,660
these are services that you can just scale vertically

40
00:01:30,660 --> 00:01:33,073
by upgrading the underlying instance type,

41
00:01:33,940 --> 00:01:35,180
although, there usually are limits

42
00:01:35,180 --> 00:01:37,710
to how much you can vertically scale

43
00:01:37,710 --> 00:01:39,280
and that's a hardware limit

44
00:01:39,280 --> 00:01:40,900
but still, vertical scalability

45
00:01:40,900 --> 00:01:44,010
is fine for a lot of use cases.

46
00:01:44,010 --> 00:01:46,570
Now, let's talk about horizontal scalability.

47
00:01:46,570 --> 00:01:47,880
Horizontal scalability means

48
00:01:47,880 --> 00:01:50,970
that you increases the number of instances/systems

49
00:01:50,970 --> 00:01:52,560
for your application.

50
00:01:52,560 --> 00:01:54,420
So, let's take again, our call center.

51
00:01:54,420 --> 00:01:57,450
We have an operator and he is being overloaded.

52
00:01:57,450 --> 00:01:59,150
I don't want to vertically scale it,

53
00:01:59,150 --> 00:02:01,250
I want to hire a second operator

54
00:02:01,250 --> 00:02:02,900
and now, I've just doubled my capacity.

55
00:02:02,900 --> 00:02:05,140
Actually, I'll hire a third operator.

56
00:02:05,140 --> 00:02:05,973
You know what?

57
00:02:05,973 --> 00:02:07,600
I'll hire six operators.

58
00:02:07,600 --> 00:02:11,020
I have horizontally scaled my call center.

59
00:02:11,020 --> 00:02:12,770
So, when you have horizontal scaling,

60
00:02:12,770 --> 00:02:15,310
that implies you have distributed systems

61
00:02:15,310 --> 00:02:17,830
and this is quite common when you have a web application

62
00:02:17,830 --> 00:02:19,130
or a modern application

63
00:02:19,130 --> 00:02:19,963
but remember,

64
00:02:19,963 --> 00:02:23,280
that not every application can be a distributed system.

65
00:02:23,280 --> 00:02:26,960
And I think it's easy, now a days, to horizontally scale,

66
00:02:26,960 --> 00:02:29,850
thanks to the cloud offerings such as Amazon EC2

67
00:02:29,850 --> 00:02:33,003
because we just right click on the web page and boom,

68
00:02:34,130 --> 00:02:37,140
all of the sudden we have a new EC2 instance

69
00:02:37,140 --> 00:02:40,040
and we can just horizontally scale our application.

70
00:02:40,040 --> 00:02:42,020
Now let's talk about high availability.

71
00:02:42,020 --> 00:02:43,720
High availability, that goes usually hand in hand

72
00:02:43,720 --> 00:02:46,270
with horizontal scaling but not all the time.

73
00:02:46,270 --> 00:02:47,430
High availability, that means

74
00:02:47,430 --> 00:02:49,530
that you're running your application or system

75
00:02:49,530 --> 00:02:51,530
in at least two data centers,

76
00:02:51,530 --> 00:02:54,410
or two availability zones in AWS.

77
00:02:54,410 --> 00:02:56,320
And he goal of high availability

78
00:02:56,320 --> 00:02:58,690
is to be able to survive a data center loss,

79
00:02:58,690 --> 00:03:02,320
so in case one center goes down, then we're still running.

80
00:03:02,320 --> 00:03:04,430
So, let's talk about our phone operators.

81
00:03:04,430 --> 00:03:06,240
Maybe I'll have three of my phone operators

82
00:03:06,240 --> 00:03:07,820
in the first building in New York

83
00:03:07,820 --> 00:03:10,356
and maybe I'll have three of my phone operators

84
00:03:10,356 --> 00:03:11,189
in the second building

85
00:03:11,189 --> 00:03:14,230
on the other side of the United States, in San Francisco.

86
00:03:14,230 --> 00:03:16,130
Now, if my building in New York

87
00:03:16,130 --> 00:03:19,720
loses their internet connection or their call connection,

88
00:03:19,720 --> 00:03:21,640
then that's okay, they can't work

89
00:03:21,640 --> 00:03:23,520
but my second building in San Francisco

90
00:03:23,520 --> 00:03:26,370
is still fine and they can still take the phone calls.

91
00:03:26,370 --> 00:03:30,250
So, in that case, my call center is highly available.

92
00:03:30,250 --> 00:03:32,720
Now, high availability, can also be passive,

93
00:03:32,720 --> 00:03:36,150
for example, when we have RDS Multi AZ, for example,

94
00:03:36,150 --> 00:03:38,870
we have a passive kind of high availability

95
00:03:38,870 --> 00:03:40,550
but it can also be active

96
00:03:40,550 --> 00:03:42,600
and this is when we have a horizontal scaling,

97
00:03:42,600 --> 00:03:43,660
so this is where, for example,

98
00:03:43,660 --> 00:03:46,310
I have all my phone calls in two buildings in New York.

99
00:03:46,310 --> 00:03:48,190
They're all taking calls at the same time.

100
00:03:48,190 --> 00:03:49,890
So, for EC2, what does that mean?

101
00:03:49,890 --> 00:03:51,240
Well, vertical scaling again,

102
00:03:51,240 --> 00:03:52,900
increasing the instance size,

103
00:03:52,900 --> 00:03:54,740
it's going to scale up or down.

104
00:03:54,740 --> 00:03:56,960
So, for example, the smallest kind of instance you can get

105
00:03:56,960 --> 00:04:01,960
in AWS today is t2.nano and this is 0.5 gigs of RAM, 1 vCPU

106
00:04:02,160 --> 00:04:05,730
and the biggest is A u-t12tb1.metal,

107
00:04:05,730 --> 00:04:10,730
which has 12.3 terabytes of Ram and 450 vCPUs,

108
00:04:10,890 --> 00:04:12,210
so quite a big instance

109
00:04:12,210 --> 00:04:13,920
and I'm sure these things will get bigger

110
00:04:13,920 --> 00:04:15,130
as time goes along.

111
00:04:15,130 --> 00:04:17,640
So, you can vertically scale from something very, very small

112
00:04:17,640 --> 00:04:19,149
to something extremely large.

113
00:04:19,149 --> 00:04:21,519
Horizontal scaling, that means you increase the number

114
00:04:21,519 --> 00:04:22,760
of instances you have

115
00:04:22,760 --> 00:04:26,170
and in AWS terms, it's called scale out or scale in.

116
00:04:26,170 --> 00:04:28,600
Out, when you increase the number of instances.

117
00:04:28,600 --> 00:04:31,120
In, when you decrease number of instances

118
00:04:31,120 --> 00:04:32,180
and this could be used

119
00:04:32,180 --> 00:04:34,480
for other scaling groups or load balancers.

120
00:04:34,480 --> 00:04:36,020
And then finally, high availability

121
00:04:36,020 --> 00:04:38,220
is when you run the same instance

122
00:04:38,220 --> 00:04:40,590
of the same application across multiple AZ

123
00:04:40,590 --> 00:04:42,340
and so this is for an auto scaler group

124
00:04:42,340 --> 00:04:43,810
that has multi AZ enabled

125
00:04:43,810 --> 00:04:47,100
or a load balancer that has multi AZ enabled, as well.

126
00:04:47,100 --> 00:04:47,933
So, that's it.

127
00:04:47,933 --> 00:04:49,580
Just a quick run down were fine

128
00:04:49,580 --> 00:04:51,880
on the terms high availability and scalability.

129
00:04:51,880 --> 00:04:53,370
They're necessary for you to understand

130
00:04:53,370 --> 00:04:55,260
when you look at the exam questions

131
00:04:55,260 --> 00:04:56,650
'cause they can trick you sometime,

132
00:04:56,650 --> 00:04:58,470
so make sure you're very confident with those.

133
00:04:58,470 --> 00:05:00,040
They're pretty easy when you think about it.

134
00:05:00,040 --> 00:05:02,010
Remember the call center in your mind

135
00:05:02,010 --> 00:05:03,360
when you have these questions.

136
00:05:03,360 --> 00:05:04,420
Okay, that's good.

137
00:05:04,420 --> 00:05:06,120
I will see yo in the next lecture.