0
1
00:00:00,150 --> 00:00:08,690
Bikers back with I have a project that wants to host six to 10 very very lightweight sorry six to 10
1

2
00:00:08,710 --> 00:00:11,020
very very lightweight Web sites.
2

3
00:00:11,020 --> 00:00:17,770
I could make a container for each but could also edit Apache config to multi host out of one container
3

4
00:00:17,820 --> 00:00:19,510
are the tradeoffs.
4

5
00:00:19,600 --> 00:00:19,900
All right.
5

6
00:00:19,930 --> 00:00:21,270
So this is a great question.
6

7
00:00:21,280 --> 00:00:28,750
Biker 2010 what you're asking is Apache servers because we've had Apache for decades and decades.
7

8
00:00:28,750 --> 00:00:34,540
Apache Web servers have the ability to run many Web sites all in one daemon.
8

9
00:00:34,600 --> 00:00:34,870
Right.
9

10
00:00:34,870 --> 00:00:41,410
So one Apache HPD could run many different Web sites that are unrelated.
10

11
00:00:41,410 --> 00:00:49,480
And should you do that vs. many containers so maybe one Web site per container or one large container
11

12
00:00:49,480 --> 00:00:51,460
image that has all the Web sites.
12

13
00:00:51,460 --> 00:00:52,600
Me my preference.
13

14
00:00:52,600 --> 00:00:58,840
If I'm running containers especially if I'm running a orchestrator like swarm or Kubernetes is always
14

15
00:00:58,840 --> 00:01:00,080
more containers.
15

16
00:01:00,100 --> 00:01:06,610
The reason is is that Apache Apache obviously has that capability but that better capability was built
16

17
00:01:06,610 --> 00:01:11,570
into Apache decades ago way before we had containers and isolation.
17

18
00:01:11,620 --> 00:01:14,890
And if I can isolate my web sites into different containers.
18

19
00:01:14,890 --> 00:01:20,560
Now in theory that means we are running more Apache demons and that's a little bit less efficient.
19

20
00:01:20,740 --> 00:01:25,810
If you're really focused on very small servers with very little resource utilization.
20

21
00:01:25,900 --> 00:01:32,740
So if you think about if you had 10 Web sites and they were all running an Apache daemon in a different
21

22
00:01:32,740 --> 00:01:38,110
container then obviously you've got a lot more executable is running that's taking more ram and potentially
22

23
00:01:38,140 --> 00:01:39,780
a little bit of wasted resources.
23

24
00:01:39,790 --> 00:01:44,350
But if you're talking six to 10 you're not going to really gain anything if you were saying I need to
24

25
00:01:44,350 --> 00:01:50,890
run a thousand Web sites and you have full control over all of them then you would have more significant
25

26
00:01:50,890 --> 00:01:56,650
resource savings by putting those all in one container as one Apache and then essentially using different
26

27
00:01:56,680 --> 00:01:59,420
Apache can figs for each Web site.
27

28
00:01:59,430 --> 00:02:03,110
Right now that's very to me that's old school that's pre container.
28

29
00:02:03,140 --> 00:02:08,400
I almost always would prefer isolating them in their own containers because that's easier to me.
29

30
00:02:08,440 --> 00:02:14,590
I use will have one repo right for the code for that Web site and I'm going to put that repo into one
30

31
00:02:14,590 --> 00:02:18,060
container image and there's that one to one relationship that makes sense to me.
31

32
00:02:18,130 --> 00:02:22,760
You can also then scale those containers independently.
32

33
00:02:22,900 --> 00:02:29,800
So if one needs to have three containers running for performance reasons or redundancy you can scale
33

34
00:02:29,800 --> 00:02:32,950
that one independently of other containers and other Web sites.
34

35
00:02:32,950 --> 00:02:38,230
Another reason if you need to update a Web site if you are doing the all in one container that means
35

36
00:02:38,230 --> 00:02:44,500
now every web server has to restart just so that you can update that one container image whereas if
36

37
00:02:44,650 --> 00:02:49,870
each one had its own container image and its own container then only the one that's changing would update.
37

38
00:02:49,870 --> 00:02:50,140
Right.
38

39
00:02:50,140 --> 00:02:54,210
So you would break potentially less you have less change overall.
39

40
00:02:54,220 --> 00:02:59,470
So anyway that's just some of the good reasons to isolate and yeah the same reason goes for a lot of
40

41
00:02:59,470 --> 00:03:01,190
other things like my sequel.
41

42
00:03:01,310 --> 00:03:03,820
You know read this stuff like that.
42

43
00:03:03,910 --> 00:03:09,880
Some people want to you know make everything into put everything on one huge Sequel Server and that's
43

44
00:03:09,880 --> 00:03:10,300
fine.
44

45
00:03:10,300 --> 00:03:16,480
That might make sense for you but there are also advantages to the isolation and the potential for updating
45

46
00:03:16,480 --> 00:03:18,340
and stuff like that of breaking these things out.
46

47
00:03:18,700 --> 00:03:24,550
So I think it's a little bit more fuzzy and not so black and white for databases because there is a
47

48
00:03:24,550 --> 00:03:29,650
lot of efficiency when you talk about memory savings and stuff when you talk about building large scale
48

49
00:03:29,650 --> 00:03:35,410
servers but when it comes to applications like web servers I always isolate them.