1
00:00:06,547 --> 00:00:08,899
- Now let's review a use case

2
00:00:08,899 --> 00:00:13,748
of video transcoding and
archival in Amazon Web Services.

3
00:00:13,748 --> 00:00:17,353
In this particular use
case, what we're looking at

4
00:00:17,353 --> 00:00:20,186
is an Amazon EC2 instance launched

5
00:00:21,338 --> 00:00:25,003
as a g2 dot two extra
large, which has access

6
00:00:25,003 --> 00:00:29,716
to some very powerful GPUs,
and we're going to use those

7
00:00:29,716 --> 00:00:32,799
to render a 3D animation to the disk,

8
00:00:33,699 --> 00:00:35,982
and because we have a lot of information

9
00:00:35,982 --> 00:00:37,791
going to the disk sequentially,

10
00:00:37,791 --> 00:00:41,476
we went with a throughput
optimized hard drive

11
00:00:41,476 --> 00:00:45,115
instead of an SSD provisioned
IOPS or general purpose.

12
00:00:45,115 --> 00:00:48,919
Now, once this animation
finishes rendering,

13
00:00:48,919 --> 00:00:52,515
what we want to do is take
that high resolution master,

14
00:00:52,515 --> 00:00:55,474
and we're going to upload that to S3.

15
00:00:55,474 --> 00:00:59,641
S3 would then be configured
to lifecycle those masters

16
00:01:00,782 --> 00:01:02,912
off to Glacier after seven days,

17
00:01:02,912 --> 00:01:04,970
so sometime within that first week,

18
00:01:04,970 --> 00:01:06,925
we might do something with the master,

19
00:01:06,925 --> 00:01:09,994
and then after that, we wanna keep it

20
00:01:09,994 --> 00:01:11,493
for a long period of time,

21
00:01:11,493 --> 00:01:13,941
but we just wanna lower
our cost of storage,

22
00:01:13,941 --> 00:01:16,596
and so we're gonna leverage
Glacier to do that.

23
00:01:16,596 --> 00:01:21,103
Now, once this object, the
high resolution master,

24
00:01:21,103 --> 00:01:24,077
has been uploaded to S3,
we're going to leverage

25
00:01:24,077 --> 00:01:28,002
S3 events to trigger a Lambda function.

26
00:01:28,002 --> 00:01:31,184
One Lambda function will take the metadata

27
00:01:31,184 --> 00:01:34,563
of that video and it
will store that metadata

28
00:01:34,563 --> 00:01:37,401
in an Amazon DynamoDB table

29
00:01:37,401 --> 00:01:39,807
so that we can reference that later,

30
00:01:39,807 --> 00:01:42,103
and we might correlate that metadata,

31
00:01:42,103 --> 00:01:44,907
one, back to the file that's stored in S3,

32
00:01:44,907 --> 00:01:47,373
and we might also correlate that metadata

33
00:01:47,373 --> 00:01:50,893
to the objects stored in Glacier as well.

34
00:01:50,893 --> 00:01:55,060
At the same time, that same
event will also trigger

35
00:01:56,061 --> 00:01:59,754
an additional Lambda
function, concurrently.

36
00:01:59,754 --> 00:02:02,592
You can see here we're
leveraging the S3 object

37
00:02:02,592 --> 00:02:06,759
created post-event to invoke
these Lambda functions.

38
00:02:07,687 --> 00:02:10,273
The second Lambda function will be used

39
00:02:10,273 --> 00:02:13,242
to trigger the Elastic Transcoder.

40
00:02:13,242 --> 00:02:15,550
The Amazon Elastic Transcoder is a service

41
00:02:15,550 --> 00:02:19,288
that we can use to transcode this video

42
00:02:19,288 --> 00:02:22,336
into a number of different formats.

43
00:02:22,336 --> 00:02:25,232
Now, again, this is the
high-resolution master,

44
00:02:25,232 --> 00:02:27,125
but we wanna get this video out there

45
00:02:27,125 --> 00:02:30,496
to people to view at a
number of different formats,

46
00:02:30,496 --> 00:02:34,363
different screen sizes,
different platforms, and so on.

47
00:02:34,363 --> 00:02:37,172
So, we're triggering
the Elastic Transcoder,

48
00:02:37,172 --> 00:02:40,827
which will pull that video from S3,

49
00:02:40,827 --> 00:02:44,143
it will transcode it into various sizes,

50
00:02:44,143 --> 00:02:45,868
it will output all of those,

51
00:02:45,868 --> 00:02:49,368
perhaps a half a dozen different videos

52
00:02:49,368 --> 00:02:51,451
to S3, to another bucket.

53
00:02:52,594 --> 00:02:56,570
We will use that bucket
as what we call the origin

54
00:02:56,570 --> 00:02:59,363
for a CloudFront distribution,

55
00:02:59,363 --> 00:03:01,296
and then we can use CloudFront

56
00:03:01,296 --> 00:03:04,225
so that these final output videos

57
00:03:04,225 --> 00:03:08,058
are served to other users
from edge locations.

58
00:03:10,141 --> 00:03:12,243
Some of these things we haven't,

59
00:03:12,243 --> 00:03:14,326
perhaps haven't talked about yet,

60
00:03:14,326 --> 00:03:17,979
and again, as far as these
use cases and demos go,

61
00:03:17,979 --> 00:03:20,415
I want to spur your imagination.

62
00:03:20,415 --> 00:03:22,991
I want to get your thinking
about the possibilities.

63
00:03:22,991 --> 00:03:26,371
I want you thinking about
ways that you can leverage

64
00:03:26,371 --> 00:03:29,814
these various services,
perhaps in similar ways

65
00:03:29,814 --> 00:03:33,674
to solve the problems that you're facing

66
00:03:33,674 --> 00:03:35,986
in your particular business.

67
00:03:35,986 --> 00:03:38,319
So, that is just one example

68
00:03:40,053 --> 00:03:43,330
of the virtually limitless possibilities

69
00:03:43,330 --> 00:03:45,961
that we have to put
these services together

70
00:03:45,961 --> 00:03:50,195
within Amazon Web Services,
to solve technical challenges,

71
00:03:50,195 --> 00:03:53,448
and that is video transcoding
and archival within,

72
00:03:53,448 --> 00:03:56,932
using the various storage
options we have available.