1
00:00:02,210 --> 00:00:08,450
Here's example number one, we had a kind of related example earlier in this course already. Let's

2
00:00:08,450 --> 00:00:15,470
say we're creating our database, our application for a hospital and there, we got patients and there, disease

3
00:00:15,560 --> 00:00:16,760
summary.

4
00:00:16,760 --> 00:00:22,660
So every patient has one disease summary which belongs to that patient only

5
00:00:22,760 --> 00:00:25,230
and there is only one summary per patient.

6
00:00:25,250 --> 00:00:28,110
So we're not talking about the history we had earlier in this course,

7
00:00:28,130 --> 00:00:29,630
it's just a summary,

8
00:00:29,640 --> 00:00:36,470
something like is ill, had previous diseases and a list of previous diseases, things like that.

9
00:00:36,470 --> 00:00:38,160
So one summary per patient

10
00:00:38,270 --> 00:00:43,640
and of course this is a one-to-one relation because the summary of patient A can never belong to patient

11
00:00:43,640 --> 00:00:44,160
B

12
00:00:44,390 --> 00:00:52,940
and the other way around, how could we model that? Back in a brand new cleaned up database server, I'll

13
00:00:52,950 --> 00:00:57,040
switch to my hospital database which will be created on the fly

14
00:00:57,470 --> 00:01:04,530
and there, I want to create my patients collection or to be precise, I want to insert a patient, so let's insert

15
00:01:04,640 --> 00:01:06,130
one patient.

16
00:01:06,240 --> 00:01:14,050
Now let's say a patient simply has a name Max, an age and then we have that diseaseSummary,

17
00:01:14,060 --> 00:01:23,390
now let's go for the reference approach first, diseaseSummary could be summary-max-1, any identifier

18
00:01:23,390 --> 00:01:25,480
you want,

19
00:01:25,480 --> 00:01:28,480
so here I'll go for that ID approach.

20
00:01:28,690 --> 00:01:29,790
We can insert that

21
00:01:29,860 --> 00:01:38,920
and obviously if I now inspect my patients with findOne pretty, we see that one patient in there, without

22
00:01:38,920 --> 00:01:41,240
pretty though, that is not existing there,

23
00:01:41,290 --> 00:01:43,940
so we see the one patient in there.

24
00:01:43,960 --> 00:01:52,630
Now let's also create another collection, diseaseSummaries and then there I'll also insert one summary,

25
00:01:52,800 --> 00:01:59,740
there I'll explicitly define the ID and use my own ID, summary-max-1, I could have also gone with the

26
00:01:59,770 --> 00:02:01,790
automatically generated objectId

27
00:02:01,930 --> 00:02:06,730
but then I should have generated the summaries collection in that entry first so that I could take

28
00:02:06,730 --> 00:02:10,150
that objectId and use it as an ID on the patient.

29
00:02:10,150 --> 00:02:11,460
I did it the other way around,

30
00:02:11,500 --> 00:02:13,870
so I'm using my own ID here

31
00:02:14,140 --> 00:02:21,890
and there I'll simply add my diseases array which simply lists all diseases

32
00:02:21,910 --> 00:02:27,430
I ever had and let's say that's a cold and that's a broken leg which I never had

33
00:02:27,430 --> 00:02:37,320
but anyways. So now I added this too and now of course I can access my diseaseSummary, summaries collection here,

34
00:02:38,280 --> 00:02:43,170
find the one element that is in there and this is the element I just inserted with my own ID and

35
00:02:43,170 --> 00:02:48,040
that ID of course matches the ID I stored as a reference on my patient.

36
00:02:48,250 --> 00:02:53,180
Now let's say in our application, the website we're building, the mobile app, whatever it is,

37
00:02:53,340 --> 00:03:00,690
we now have a request where we need that patient and also the disease history or the disease summary I should

38
00:03:00,690 --> 00:03:01,360
say.

39
00:03:01,860 --> 00:03:08,160
So we might have an incoming request which in the end by the mongodb driver is translated to a

40
00:03:08,160 --> 00:03:09,800
find request,

41
00:03:09,810 --> 00:03:16,770
so we first of all find our patients and in reality, we'll probably do that by looking for let's say

42
00:03:16,770 --> 00:03:23,370
all patients with a name of Max or here as a shortcut since I only have one, I'll just use findOne,

43
00:03:23,370 --> 00:03:27,310
the thing is in the end we have some query that returns Max here.

44
00:03:27,780 --> 00:03:30,420
So that is my returned 

45
00:03:30,600 --> 00:03:31,610
patient data,

46
00:03:31,650 --> 00:03:34,750
the problem is I'm also interested in the diseaseSummary,

47
00:03:34,890 --> 00:03:37,000
so I also want to get that.

48
00:03:37,230 --> 00:03:46,350
Now what I can of course do is in my application, I can access diseaseSummary and therefore get my key

49
00:03:46,350 --> 00:03:54,180
here and in the application we're writing, in PHP, .NET, Java, NodeJS, whatever it is, we can of course

50
00:03:54,300 --> 00:03:58,120
fetch some data from a server and then store that data in a variable,

51
00:03:58,140 --> 00:04:07,240
so let's say here in the shell, we can do this with the var keyword, we have our dsid for diseaseSummary

52
00:04:07,240 --> 00:04:15,280
ID and that simply is the result of patients findOne and then there, accessing the diseaseSummary,

53
00:04:15,460 --> 00:04:24,100
so now if we output dsid, that is the ID stored in this patient. And now we can simply reach out to

54
00:04:24,100 --> 00:04:27,130
diseaseSummaries, so that other collection

55
00:04:27,370 --> 00:04:37,800
and there we want to find the first diseaseSummary where _id is equal to dsid,

56
00:04:37,830 --> 00:04:39,630
so I can use that variable here

57
00:04:39,870 --> 00:04:45,810
and I simply stored it in a variable so that I don't have to paste or add this long term here as a value

58
00:04:45,810 --> 00:04:46,580
for ID.

59
00:04:46,920 --> 00:04:49,640
If I do that, I indeed find my diseaseSummary,

60
00:04:49,650 --> 00:04:52,420
the bad thing here is that this took two steps

61
00:04:52,530 --> 00:04:56,060
and also from a performance perspective, this will actually be super fast

62
00:04:56,070 --> 00:05:01,640
but if you have a very big database, a very big collection of patients and a big collection of

63
00:05:01,640 --> 00:05:02,580
diseaseSummaries,

64
00:05:02,580 --> 00:05:08,880
this might not be the optimal way of doing that, splitting that across two collections feels a bit unnecessary

65
00:05:08,880 --> 00:05:09,870
here,

66
00:05:09,870 --> 00:05:15,380
the better approach in such a case where we have a strong one-to-one relation would be to use embedded

67
00:05:15,380 --> 00:05:16,780
document.

68
00:05:16,800 --> 00:05:24,660
so let me actually reach out to my patients and delete all patients in there

69
00:05:24,930 --> 00:05:27,440
and now let's insert that patient again,

70
00:05:27,570 --> 00:05:31,370
so that insert one command from earlier where I inserted Max,

71
00:05:31,560 --> 00:05:38,480
now I will not insert a diseaseSummary which is a reference but instead I'll insert a nested, an embedded

72
00:05:38,490 --> 00:05:47,400
document here and now that embedded document would have my diseases key, my diseases field with the

73
00:05:47,400 --> 00:05:51,500
cold and the broken leg, like this.

74
00:05:51,720 --> 00:05:56,910
And now the huge advantage of course is that in the application we're writing with whichever language

75
00:05:56,910 --> 00:06:05,760
we use, I can simply write my query to find my patient by ID, by name or in this case as a shortcut, simply

76
00:06:05,760 --> 00:06:09,060
with findOne and I will get all the data I need,

77
00:06:09,060 --> 00:06:11,850
the patient data and the diseaseSummary data.

78
00:06:12,060 --> 00:06:17,870
So this is an example for when you would use an embed document, you have a strong one-to-one relationship.