1
00:00:06,451 --> 00:00:08,670
- Now let's take a look at an overview

2
00:00:08,670 --> 00:00:11,120
of all the options we have for storage

3
00:00:11,120 --> 00:00:13,370
within Amazon Web Services.

4
00:00:14,829 --> 00:00:16,778
We have a number of
different storage options

5
00:00:16,778 --> 00:00:18,550
in Amazon Web Services,

6
00:00:18,550 --> 00:00:19,989
and let's just start here on the left

7
00:00:19,989 --> 00:00:23,067
with what we call the instance store.

8
00:00:23,067 --> 00:00:25,938
The instance store refers to

9
00:00:25,938 --> 00:00:30,488
block storage that are,
essentially what we could say is,

10
00:00:30,488 --> 00:00:33,890
are built in to the EC2 instance.

11
00:00:33,890 --> 00:00:36,417
A number of different EC2 instance types

12
00:00:36,417 --> 00:00:39,884
come with a different type of storage.

13
00:00:39,884 --> 00:00:42,336
It could be a small 30 gig volume,

14
00:00:42,336 --> 00:00:45,840
it could be several
terabytes worth of volumes,

15
00:00:45,840 --> 00:00:50,459
and these are, now we say, a
free, sort of highlighted here

16
00:00:50,459 --> 00:00:53,378
because the cost of
that storage is included

17
00:00:53,378 --> 00:00:54,798
with the price of the machine,

18
00:00:54,798 --> 00:00:57,378
so you're not paying anything additional

19
00:00:57,378 --> 00:00:59,377
to the price of the
machine for that storage,

20
00:00:59,377 --> 00:01:01,147
it just comes with it.

21
00:01:01,147 --> 00:01:05,307
Now, these drives within
the instance store

22
00:01:05,307 --> 00:01:07,138
are considered ephemeral,

23
00:01:07,138 --> 00:01:12,036
meaning that when that EC2
instance terminates, or stops,

24
00:01:12,036 --> 00:01:15,470
then whatever data is on that
particular instance store

25
00:01:15,470 --> 00:01:16,719
will be gone.

26
00:01:16,719 --> 00:01:19,410
And that's because the instance store

27
00:01:19,410 --> 00:01:22,637
is backed by the volumes that are present

28
00:01:22,637 --> 00:01:25,949
on the physical host of
that virtual machine.

29
00:01:25,949 --> 00:01:27,705
So, when you stop a virtual machine,

30
00:01:27,705 --> 00:01:29,255
or terminate a virtual machine,

31
00:01:29,255 --> 00:01:31,022
those stores will go away.

32
00:01:31,022 --> 00:01:33,851
Even if you were to stop a
machine and start it again,

33
00:01:33,851 --> 00:01:37,852
it's most likely, probably
almost even guaranteed

34
00:01:37,852 --> 00:01:40,441
that that instance will come back

35
00:01:40,441 --> 00:01:42,444
on a different physical host

36
00:01:42,444 --> 00:01:44,899
and so that underlining storage subsystem

37
00:01:44,899 --> 00:01:46,449
is no longer available.

38
00:01:46,449 --> 00:01:50,159
And so, that's why we consider
the instance store ephemeral.

39
00:01:50,159 --> 00:01:51,507
It's really important to know that

40
00:01:51,507 --> 00:01:53,148
so if that if you have applications

41
00:01:53,148 --> 00:01:54,842
that are writing to that,

42
00:01:54,842 --> 00:01:57,763
then if you lose that
machine for whatever reason,

43
00:01:57,763 --> 00:01:59,863
if it fails, if you terminate it,

44
00:01:59,863 --> 00:02:02,404
if auto-scaling pulls that machine back in

45
00:02:02,404 --> 00:02:03,564
and gets rid of it,

46
00:02:03,564 --> 00:02:05,535
then that storage will go away.

47
00:02:05,535 --> 00:02:08,525
We do not have the
ability to take snapshots,

48
00:02:08,525 --> 00:02:12,314
so there's no snapshot
capabilities of the instance store.

49
00:02:12,314 --> 00:02:16,245
If you want to use that for
it's high IOPs capability,

50
00:02:16,245 --> 00:02:17,955
then back ups will be something

51
00:02:17,955 --> 00:02:20,374
that you will have to do on your own.

52
00:02:20,374 --> 00:02:23,414
Now, that brings us to
another form of block storage,

53
00:02:23,414 --> 00:02:27,219
which is Amazon EBS, Elastic
Block Storage service.

54
00:02:27,219 --> 00:02:29,674
Now, the nice thing
about the service is that

55
00:02:29,674 --> 00:02:31,772
these volumes, these disks,

56
00:02:31,772 --> 00:02:34,886
are independent from the machine itself.

57
00:02:34,886 --> 00:02:38,047
Typically what we would do
is launch an EC2 instance

58
00:02:38,047 --> 00:02:42,214
and then we would attach an
EBS volume to that instance

59
00:02:43,355 --> 00:02:46,543
so that we can have a
place to store our data.

60
00:02:46,543 --> 00:02:50,710
But, because this device is
independent from the machine,

61
00:02:51,947 --> 00:02:54,527
if we configure things properly,

62
00:02:54,527 --> 00:02:57,790
we could potentially, you
know, lose this EC2 instance,

63
00:02:57,790 --> 00:03:01,177
but the data that's on this EBS volume

64
00:03:01,177 --> 00:03:02,695
will still be there

65
00:03:02,695 --> 00:03:06,256
because it's independent from
the life of the EC2 instance.

66
00:03:06,256 --> 00:03:08,923
Now, it's important to know that

67
00:03:10,296 --> 00:03:12,494
as far as EBS goes,

68
00:03:12,494 --> 00:03:16,577
what we pay for is a
allocated storage per month,

69
00:03:18,035 --> 00:03:21,952
so if we allocate one
terabyte for that volume,

70
00:03:24,090 --> 00:03:26,655
but we only use one meg,

71
00:03:26,655 --> 00:03:28,262
we're still paying for the one terabyte

72
00:03:28,262 --> 00:03:32,979
because that's how much we've
allocated for the month,

73
00:03:32,979 --> 00:03:36,668
or that's how much we've
allocated for that drive.

74
00:03:36,668 --> 00:03:39,597
Now, with EBS volumes we get

75
00:03:39,597 --> 00:03:42,847
a maximum of a 16 terabytes per volume,

76
00:03:43,850 --> 00:03:46,028
and when we say that they're durable,

77
00:03:46,028 --> 00:03:47,111
it means that

78
00:03:47,949 --> 00:03:51,699
the EBS system, inherently
behind the scenes will

79
00:03:51,699 --> 00:03:55,968
replicate the data to numerous
devices within a cluster

80
00:03:55,968 --> 00:03:57,908
in one availability zone.

81
00:03:57,908 --> 00:04:01,152
So EBS is, to some
degree, already redundant

82
00:04:01,152 --> 00:04:03,500
and it's resilient to the loss of a device

83
00:04:03,500 --> 00:04:04,920
within that cluster.

84
00:04:04,920 --> 00:04:09,087
EBS is not resilient to
the loss of a data center,

85
00:04:09,996 --> 00:04:13,818
EBS volumes are created
per availability zone,

86
00:04:13,818 --> 00:04:16,394
they don't span the entire region.

87
00:04:16,394 --> 00:04:20,927
So, we do have the ability
to create snapshots,

88
00:04:20,927 --> 00:04:25,094
and those snapshots are
then stored over here in S3.

89
00:04:27,103 --> 00:04:30,939
So, S3 is what we would
consider object storage,

90
00:04:30,939 --> 00:04:33,375
and we're gonna talk about the
difference between those two

91
00:04:33,375 --> 00:04:35,556
here in just a little bit.

92
00:04:35,556 --> 00:04:38,139
So, with S3 we can get storage,

93
00:04:39,021 --> 00:04:40,830
at least at the time of this video,

94
00:04:40,830 --> 00:04:44,223
storage was somewhere around
3 cents per gig per month.

95
00:04:44,223 --> 00:04:47,269
That's an incredibly
low price for storage,

96
00:04:47,269 --> 00:04:49,498
and what we might describe

97
00:04:49,498 --> 00:04:53,665
Amazon's Simple Storage Service
as write once, read many,

98
00:04:55,280 --> 00:05:00,211
we might also describe S3
as storage for the internet.

99
00:05:00,211 --> 00:05:04,048
So, we say that because
S3 is a really great place

100
00:05:04,048 --> 00:05:06,837
to store things that need to be retrieved

101
00:05:06,837 --> 00:05:10,745
from millions of users out
there across the internet.

102
00:05:10,745 --> 00:05:13,512
So we could put images that
are meant for a website,

103
00:05:13,512 --> 00:05:16,012
we could put software patches,

104
00:05:16,985 --> 00:05:19,463
PDF documents, videos,

105
00:05:19,463 --> 00:05:21,843
all kinds of things can be stored in S3,

106
00:05:21,843 --> 00:05:25,143
and then we can send our
users directly to S3,

107
00:05:25,143 --> 00:05:26,800
bypassing our own systems

108
00:05:26,800 --> 00:05:28,640
and reducing the load on our systems

109
00:05:28,640 --> 00:05:31,290
and just letting S3 handle the delivery

110
00:05:31,290 --> 00:05:33,928
of that content to our users.

111
00:05:33,928 --> 00:05:37,469
It's also important to know
that all transfers use HTTP,

112
00:05:37,469 --> 00:05:42,136
that is the protocol for
Amazon Simple Storage Service.

113
00:05:42,136 --> 00:05:44,156
Now, the last option
that I wanna talk about

114
00:05:44,156 --> 00:05:46,180
with AWS storage options

115
00:05:46,180 --> 00:05:48,898
is what's called the AWS Storage Gateway.

116
00:05:48,898 --> 00:05:51,912
This is a really
interesting service in that

117
00:05:51,912 --> 00:05:56,307
ultimately what it is
is a virtual machine,

118
00:05:56,307 --> 00:05:58,687
this is published by Amazon Web Services,

119
00:05:58,687 --> 00:06:02,128
and it's designed to be run on-premises,

120
00:06:02,128 --> 00:06:04,029
on an on-premises appliance

121
00:06:04,029 --> 00:06:06,112
using VM wear or Hyper V,

122
00:06:07,068 --> 00:06:08,557
and so you run this in your own

123
00:06:08,557 --> 00:06:12,786
local on-premises data center
or co-location, what have you.

124
00:06:12,786 --> 00:06:14,703
And it exposes a device

125
00:06:16,356 --> 00:06:19,427
to which you can use as local storage.

126
00:06:19,427 --> 00:06:22,106
All of the data that is stored there

127
00:06:22,106 --> 00:06:26,845
will be also backed up,
encrypted in S3 or Glacier.

128
00:06:26,845 --> 00:06:28,724
You can configure it

129
00:06:28,724 --> 00:06:30,945
to use either one of those services.

130
00:06:30,945 --> 00:06:32,733
If you use S3,

131
00:06:32,733 --> 00:06:36,933
we can use some other mechanisms
that we'll talk about,

132
00:06:36,933 --> 00:06:40,213
like life cycle rules to get those files

133
00:06:40,213 --> 00:06:42,482
from S3 to Glacier in another way.

134
00:06:42,482 --> 00:06:46,804
The storage gateway comes in
three possible configurations.

135
00:06:46,804 --> 00:06:50,014
One of them that we call
the gateway-cached volume,

136
00:06:50,014 --> 00:06:52,383
means that the great majority of your data

137
00:06:52,383 --> 00:06:54,684
is stored directly in S3,

138
00:06:54,684 --> 00:06:57,706
but frequently-accessed
data is stored locally,

139
00:06:57,706 --> 00:07:00,306
that's the gateway-cached volume.

140
00:07:00,306 --> 00:07:02,099
The gateway-stored volume

141
00:07:02,099 --> 00:07:06,107
means that we're storing
all of our data locally,

142
00:07:06,107 --> 00:07:08,889
but it takes regular
point-in-time snapshots

143
00:07:08,889 --> 00:07:11,989
and uploads those snapshots to S3,

144
00:07:11,989 --> 00:07:13,910
and from there, if we wanted to,

145
00:07:13,910 --> 00:07:17,827
we could rebuild those
snapshots as EBS volumes

146
00:07:19,158 --> 00:07:23,055
and attach those to virtual
machines within EC2.

147
00:07:23,055 --> 00:07:26,774
The last one is what's
called a virtual tape library

148
00:07:26,774 --> 00:07:29,143
that exposes an iSCSI interface,

149
00:07:29,143 --> 00:07:31,692
and it essentially looks just like tape,

150
00:07:31,692 --> 00:07:34,822
but that each virtual tape is stored

151
00:07:34,822 --> 00:07:37,213
directly in S3, or Glacier.

152
00:07:37,213 --> 00:07:39,929
So, for those instances where you have

153
00:07:39,929 --> 00:07:41,779
a lot of data on premises

154
00:07:41,779 --> 00:07:43,449
and you have a lot of heavy workload,

155
00:07:43,449 --> 00:07:47,139
you're not ready to move all
of that stuff to the Cloud yet,

156
00:07:47,139 --> 00:07:49,130
or perhaps there are things that you

157
00:07:49,130 --> 00:07:50,741
never plan to migrate,

158
00:07:50,741 --> 00:07:52,530
you're gonna leave it back on premises,

159
00:07:52,530 --> 00:07:54,410
but you want a good DR solution,

160
00:07:54,410 --> 00:07:56,387
then the AWS Storage Gateway

161
00:07:56,387 --> 00:07:58,636
is a really good option to consider.

162
00:07:58,636 --> 00:08:02,519
So, those at a high level,
those are our storage options

163
00:08:02,519 --> 00:08:04,769
within Amazon Web Services.