1
00:00:00,300 --> 00:00:02,070
‫So, now let's talk about S3 Select

2
00:00:02,070 --> 00:00:03,360
‫and Glacier Select.

3
00:00:03,360 --> 00:00:05,340
‫The idea is that you know that you want to retrieve

4
00:00:05,340 --> 00:00:06,720
‫a file from S3,

5
00:00:06,720 --> 00:00:09,090
‫but then you're going to filter it after retrieving it,

6
00:00:09,090 --> 00:00:11,730
‫and therefore you're retrieving too much data.

7
00:00:11,730 --> 00:00:13,440
‫What if instead you could use SQL

8
00:00:13,440 --> 00:00:15,390
‫to perform server-side filtering?

9
00:00:15,390 --> 00:00:17,370
‫Therefore, that means that you're going to filter

10
00:00:17,370 --> 00:00:20,130
‫by rows or by columns using simple SQL statements

11
00:00:20,130 --> 00:00:22,410
‫to have less network transfers

12
00:00:22,410 --> 00:00:23,940
‫and less CPU cost client-side

13
00:00:23,940 --> 00:00:26,790
‫to actually go through the data and filter it.

14
00:00:26,790 --> 00:00:28,380
‫So before S3 Select,

15
00:00:28,380 --> 00:00:30,810
‫what you do is that you retrieve all the data

16
00:00:30,810 --> 00:00:33,060
‫and then, application-side, you're going to filter it

17
00:00:33,060 --> 00:00:34,320
‫to find what you need.

18
00:00:34,320 --> 00:00:35,730
‫And that's a lot of data coming in

19
00:00:35,730 --> 00:00:37,950
‫for only a little bit of data being used.

20
00:00:37,950 --> 00:00:39,510
‫But if you use S3 Select,

21
00:00:39,510 --> 00:00:42,960
‫you actually have Amazon S3 filter the file for you,

22
00:00:42,960 --> 00:00:44,820
‫and you only retrieve the data you need.

23
00:00:44,820 --> 00:00:47,910
‫Therefore, Amazon claims it's up to 400% faster

24
00:00:47,910 --> 00:00:51,150
‫and 80% cheaper to use S3 Select.

25
00:00:51,150 --> 00:00:52,680
‫Put it in another diagram,

26
00:00:52,680 --> 00:00:54,450
‫we're going to want to get a CSV

27
00:00:54,450 --> 00:00:56,610
‫with Amazon S3 Select, okay.

28
00:00:56,610 --> 00:00:59,070
‫Amazon S3 is going to find the CSV file

29
00:00:59,070 --> 00:01:02,310
‫and filter it server-side, so on its own service,

30
00:01:02,310 --> 00:01:04,680
‫and then it will send us the filtered data sets

31
00:01:04,680 --> 00:01:07,770
‫so that it's much smaller and much cheaper.

32
00:01:07,770 --> 00:01:11,070
‫So for simple filtering, think about S3 Select

33
00:01:11,070 --> 00:01:14,430
‫and that also works on Glacier, so Glacier Select.

34
00:01:14,430 --> 00:01:16,730
‫That's it. I will see you in the next lecture.