1
00:00:00,060 --> 00:00:03,610
‫Horizontal partitioning or just partitioning versus SHARDEY?

2
00:00:03,900 --> 00:00:09,300
‫So, guys, we made a video about sharding partitioning, but the story is very interesting and let's

3
00:00:09,300 --> 00:00:12,260
‫discuss the difference between the two horizontal partition.

4
00:00:12,270 --> 00:00:18,120
‫We just partition splits the big table into multiple tables in the same database.

5
00:00:18,540 --> 00:00:25,880
‫And the database takes care of the management of these partition you as a client, as a query builder.

6
00:00:26,250 --> 00:00:28,500
‫You have no idea what partition you're hitting.

7
00:00:28,680 --> 00:00:30,050
‫The database is designing.

8
00:00:30,060 --> 00:00:32,610
‫So that's why I said the client is agnostic.

9
00:00:32,790 --> 00:00:35,840
‫It doesn't care which partition is going to hit the database.

10
00:00:36,030 --> 00:00:41,910
‫I'm sending my own beautiful query and the database decides which partition to head based on the logic

11
00:00:41,910 --> 00:00:42,750
‫on the where clause.

12
00:00:43,240 --> 00:00:44,310
‫So that's the trickier.

13
00:00:44,940 --> 00:00:49,710
‫Sharding, on the other hand, splits the big table into all the multiple tables.

14
00:00:49,800 --> 00:00:53,220
‫However, they are completely different servers.

15
00:00:53,250 --> 00:00:54,960
‫They are on a completely different database server.

16
00:00:55,080 --> 00:00:59,340
‫And you do this for for distributed essentially processing.

17
00:00:59,340 --> 00:00:59,600
‫Right.

18
00:00:59,740 --> 00:01:02,610
‫So like, let's throw this into different tables.

19
00:01:02,610 --> 00:01:02,850
‫Right.

20
00:01:02,850 --> 00:01:05,100
‫All together in different servers altogether.

21
00:01:05,460 --> 00:01:13,110
‫So that like basically for latency issues, like I say, California customers, we're going to put a

22
00:01:13,110 --> 00:01:16,980
‫database, Shahd, just for California customers.

23
00:01:17,130 --> 00:01:23,520
‫And you're going to hit if you're from California, we're going to hit you based on on a like a proxy

24
00:01:23,530 --> 00:01:24,150
‫school.

25
00:01:24,300 --> 00:01:30,360
‫We're going to detect where IP addresses are from and we're going to serve you to the database or if

26
00:01:30,360 --> 00:01:37,590
‫it's read, for example, to the database instance that has the California table.

27
00:01:38,340 --> 00:01:38,730
‫Right.

28
00:01:38,730 --> 00:01:45,140
‫Versus if you're going from Asia, we're going to hit you with an Asian server.

29
00:01:45,270 --> 00:01:45,600
‫Right.

30
00:01:45,610 --> 00:01:49,150
‫So that's the Hidatsa IP address is different, completely different IP address.

31
00:01:49,350 --> 00:01:51,270
‫So that's a beautiful, beautiful design.

32
00:01:51,450 --> 00:01:54,060
‫Obviously, it's a little bit complicated, but but yeah.

33
00:01:54,240 --> 00:01:58,650
‫So if you think about it, the client is aware of the of the Shard.

34
00:01:58,770 --> 00:01:59,160
‫Right.

35
00:01:59,340 --> 00:02:02,080
‫Which is Bignone on and sharding.

36
00:02:02,100 --> 00:02:04,620
‫That's that's one of the big thing that I don't like about Chani.

37
00:02:05,130 --> 00:02:09,360
‫And I know Vitesse and other technologies are trying to solve this.

38
00:02:09,360 --> 00:02:15,570
‫But still it's a it's a it's a big it's a big investment for anyone back engineers.

39
00:02:16,800 --> 00:02:22,320
‫So the table names, this is like a little bit of just detail the horizontal partitioning table name

40
00:02:22,320 --> 00:02:23,850
‫changes of that back end.

41
00:02:24,030 --> 00:02:26,670
‫You have like multiple table names and they are changed.

42
00:02:26,970 --> 00:02:27,330
‫Right.

43
00:02:27,490 --> 00:02:31,650
‫However, in the Sharding case, the timetabling stays the same almost.

44
00:02:31,650 --> 00:02:31,890
‫Right.

45
00:02:31,890 --> 00:02:36,300
‫Because the query is exactly identical.

46
00:02:36,300 --> 00:02:37,140
‫Doesn't change.

47
00:02:37,410 --> 00:02:37,770
‫Right.

48
00:02:38,040 --> 00:02:43,410
‫But the server way to which you're executing the query against actually changes.