1
00:00:02,140 --> 00:00:03,870
So what are transactions?

2
00:00:04,090 --> 00:00:09,400
Well let's consider this use case, we have a users and we have a posts collection and we get a couple

3
00:00:09,400 --> 00:00:15,460
of documents in there and let's say each user or most users have a couple of posts. So the posts are

4
00:00:15,460 --> 00:00:19,480
related to users because the user is the person who created the post.

5
00:00:19,750 --> 00:00:27,460
So we have that stored relation, maybe stored with a reference, so with a key stored in a user

6
00:00:27,460 --> 00:00:29,890
or in a post document, doesn't really matter.

7
00:00:29,890 --> 00:00:36,280
Now we delete the user account and therefore we want to delete the documents here together,

8
00:00:36,280 --> 00:00:40,660
so in two collections. Now this can be done without transactions,

9
00:00:40,660 --> 00:00:43,600
we can simply delete a user

10
00:00:44,110 --> 00:00:49,930
and right before we do that, we save the ID of that user, we find out the user id and then we reach out

11
00:00:49,930 --> 00:00:55,680
to the posts collection, find all posts who are linking to that user ID and delete these posts,

12
00:00:55,690 --> 00:00:58,930
that is perfectly possible without transactions.

13
00:00:58,930 --> 00:01:06,700
Now the thing just is what happens when we actually have a use case where deleting the user succeeds

14
00:01:07,030 --> 00:01:11,070
but during deleting the posts, somehow something goes wrong?

15
00:01:11,170 --> 00:01:13,360
We have a temporary server outage,

16
00:01:13,390 --> 00:01:15,580
we have a network issue, whatever it is.

17
00:01:15,580 --> 00:01:22,960
So then we end up in a state where the user was deleted but the posts are still there but the user they're

18
00:01:22,960 --> 00:01:25,700
pointing at doesn't exist anymore,

19
00:01:26,200 --> 00:01:30,440
this is exactly the use case where transactions help us. With a transaction,

20
00:01:30,520 --> 00:01:37,120
we can basically tell mongodb hey these operations as many as you want either succeed together

21
00:01:37,270 --> 00:01:40,680
or they fail together and you roll back,

22
00:01:40,750 --> 00:01:46,810
so you restore the database in the state it was in before the transaction regarding the documents

23
00:01:46,930 --> 00:01:49,130
that were affected in the transaction.

24
00:01:49,240 --> 00:01:51,220
That is the idea behind transactions,

25
00:01:51,220 --> 00:01:54,500
let's now try that out and again for that, you need mongodb 4

26
00:01:54,610 --> 00:01:59,510
and a replica set, so that is why I will use the mongodb Atlas deployment

27
00:01:59,530 --> 00:02:01,060
we created in the last module.