1
00:00:00,680 --> 00:00:06,230
Now, if you want to use tracking technology, you need to buy the right switches, so essentially you

2
00:00:06,230 --> 00:00:09,750
need to buy the right product for the feature that you want.

3
00:00:10,370 --> 00:00:14,060
As mentioned, Cisco have supported stacking for a long time.

4
00:00:14,810 --> 00:00:15,290
Such as, 

5
00:00:15,290 --> 00:00:18,900
such as the 3750 have supported stackwise for many years.

6
00:00:19,340 --> 00:00:20,450
Here's some examples.

7
00:00:20,990 --> 00:00:24,020
FlexStack was introduced in 2010.

8
00:00:24,620 --> 00:00:27,890
FlexStack plus was introduced in 2013.

9
00:00:28,400 --> 00:00:35,270
The switches that support FlexStack are the 2960s and 2960x for FlexStack 

10
00:00:35,270 --> 00:00:38,120
Plus you need a 2960x or 29

11
00:00:38,120 --> 00:00:42,830
60xr. The speed of a single link in both directions

12
00:00:42,980 --> 00:00:49,250
using full duplex is 10 gigabits per second for FlexStack and 20 gigabits per second for FlexStack

13
00:00:49,250 --> 00:00:56,630
plus. The maximum number of switches supported in one stack is 4 for FlexStack and 8 for Flex

14
00:00:56,630 --> 00:00:57,080
Stack

15
00:00:57,080 --> 00:00:59,540
Plus. In the real world

16
00:00:59,540 --> 00:01:05,960
have a look on the Cisco documentation and the data sheets for any switch that you want to buy to ensure

17
00:01:05,960 --> 00:01:09,800
that it supports the speeds and capabilities that you need.

18
00:01:10,520 --> 00:01:17,900
Now, Chassis aggregation is another Cisco technology which allows you to make multiple switches operate

19
00:01:17,900 --> 00:01:18,920
as a single switch.

20
00:01:19,640 --> 00:01:21,590
From a big picture perspective.

21
00:01:21,980 --> 00:01:29,180
In a lot of cases, switch stacking is used after the access layer, whereas Chessie aggregation is

22
00:01:29,180 --> 00:01:32,990
used for more powerful switches used in the distribution.

23
00:01:32,990 --> 00:01:34,110
And Callejas.

24
00:01:34,820 --> 00:01:38,750
So in summary, Chessie aggregation is used for high end switches.

25
00:01:39,140 --> 00:01:44,690
As an example, Chessie based switches used in the distribution and Callejas of campus networks.

26
00:01:45,300 --> 00:01:50,900
It does not require special hardware adapters, but rather uses Ethernet interfaces on switches.

27
00:01:51,290 --> 00:01:53,900
It typically aggregates only two switches.

28
00:01:54,440 --> 00:01:58,580
It's more complex to configure, but provides more options.

29
00:01:59,480 --> 00:02:05,060
Now, from a big picture point of view, Chessie aggregation is the same as switch stacking.

30
00:02:05,510 --> 00:02:11,870
Multiple switches act as one switch, which gives you both availability and design advantages.

31
00:02:12,620 --> 00:02:19,100
However, one of the big reasons for Chessie based aggregation is high availability designs.

32
00:02:19,580 --> 00:02:28,010
Technologies such as Cisco Virtual Switching System, or VSS is supported on the Cisco 6500 and 6800

33
00:02:28,010 --> 00:02:28,990
series switches.

34
00:02:29,990 --> 00:02:34,370
Have a look at Cisco's website for more details, but he has a quick overview.

35
00:02:35,210 --> 00:02:41,510
Now, even if you're not using Chessie aggregation, you need high availability in the core and distribution

36
00:02:41,510 --> 00:02:42,980
layer of your network.

37
00:02:43,760 --> 00:02:50,150
As discussed, one of the reasons for having multiple switches in the distribution or KOLIA is to provide

38
00:02:50,150 --> 00:02:53,050
redundancy in case one of those switches goes down.

39
00:02:53,540 --> 00:03:01,760
So we use technology such as HSP, spanning tree and others to provide better redundancy and better

40
00:03:01,760 --> 00:03:02,570
scalability.

41
00:03:03,320 --> 00:03:05,700
However, the downside is cost.

42
00:03:05,990 --> 00:03:10,360
You need additional switches and it's also more complex to configure.

43
00:03:10,370 --> 00:03:16,580
You need to think about where you put your spending tree roots as well as your HSP active rodders.

44
00:03:17,180 --> 00:03:25,010
Now, Chessie based switch typically has multiple line cords, one or more supervisor modules and one

45
00:03:25,010 --> 00:03:27,530
or more power supplies for redundancy.

46
00:03:27,530 --> 00:03:33,230
You want redundant power supplies, you want redundant supervisors and you want multiple line cords

47
00:03:33,230 --> 00:03:34,100
in your Chessie.

48
00:03:34,760 --> 00:03:40,880
The idea with supervisor modules is if one of the supervisors goes down, the other one can take over

49
00:03:40,880 --> 00:03:42,220
the management of the switch.

50
00:03:42,620 --> 00:03:47,140
A supervisor module is essentially the brain for the chassis based switch.

51
00:03:47,990 --> 00:03:50,630
If you lose your supervisor, the switch will have no brain.

52
00:03:51,020 --> 00:03:54,500
Hence you have redundant supervisor modules in your switch.

53
00:03:55,160 --> 00:03:59,480
You have redundant power supplies in case there's a problem with one of the power supplies.

54
00:03:59,840 --> 00:04:05,180
And in addition, you'll have multiple connections from your access layer to multiple line cords using

55
00:04:05,180 --> 00:04:11,330
link aggregation to ensure that if one of the line cords goes down, the network can continue functioning

56
00:04:11,330 --> 00:04:14,060
using the redundant line code.

57
00:04:14,390 --> 00:04:21,019
Now, with chassis based aggregation, what we're doing is taking multiple chassis based switches and

58
00:04:21,019 --> 00:04:29,030
using either related to a through ISO channel between multiple chassis based switches to provide better

59
00:04:29,030 --> 00:04:35,090
redundancy and better throughput to the distribution core of the network.

60
00:04:35,300 --> 00:04:41,630
I've discussed that in a lot of detail in the campus videos that make up this course.

61
00:04:42,020 --> 00:04:44,060
What we can do is take that a step further.

62
00:04:44,360 --> 00:04:50,120
And instead of using link aggregation between the chassis based switches with spanning three inches

63
00:04:50,150 --> 00:04:56,720
or P, we make the chassis based switches appear to be a single switch in the model.

64
00:04:56,720 --> 00:04:59,870
On the left, the two switches are independent of one another.

65
00:05:00,560 --> 00:05:03,500
They run their own Mac address tables.

66
00:05:03,500 --> 00:05:06,650
They run the own instance of spending three.

67
00:05:07,460 --> 00:05:10,520
They essentially act totally independently of one another.

68
00:05:10,940 --> 00:05:16,700
You configure two separate switches in this example and you configure them independently of each other

69
00:05:17,120 --> 00:05:19,250
with an aggregated chassis environment.

70
00:05:19,250 --> 00:05:24,200
However, the switches appear to be one switch to the rest of the network.

71
00:05:24,830 --> 00:05:31,610
You can have multiple physical ports going to different physical switches, but you can aggregate them

72
00:05:31,610 --> 00:05:39,470
together using multi chassis etha channel because a logically a one has two physical connections to

73
00:05:39,470 --> 00:05:46,520
the same switch, even though physically it's to physical connections to different switches and the

74
00:05:46,520 --> 00:05:48,200
different ways to implement this.

75
00:05:48,530 --> 00:05:55,670
We can use multi chassis etha channel but use an active standby control plane where one of the pairs

76
00:05:56,030 --> 00:05:59,270
acts as the switch for the control plane protocols.

77
00:05:59,870 --> 00:06:04,910
So one of the switches is in control of spending Treva Etha channel up and running protocols.

78
00:06:05,450 --> 00:06:12,260
But to take advantage of the fording power of the supervisor modules on both switches, we have active

79
00:06:12,260 --> 00:06:18,350
active data planes related to fording and layer three forwarding is done by both switches.

80
00:06:18,830 --> 00:06:23,210
The switches synchronize their macarius tables and routing tables to support the.

81
00:06:24,430 --> 00:06:29,890
There's a single switch management plan, however, in other words, you manage both switches on the

82
00:06:29,890 --> 00:06:37,780
active switch when you change the config of the active switch, that configuration is synchronized automatically

83
00:06:37,780 --> 00:06:39,100
with the standby switch.

84
00:06:39,760 --> 00:06:46,540
Now, you could take this a step further where you have an aggregated virtual switch and an aggregated

85
00:06:46,660 --> 00:06:47,590
access switch.

86
00:06:47,980 --> 00:06:55,150
Physically, we've got to two switches in the distribution layer, but by using Chessie aggregation,

87
00:06:55,150 --> 00:06:58,740
they appear to be one switch to the axis layer.

88
00:06:58,750 --> 00:07:04,930
We have four physical switches, but they appear to be a single virtual switch.

89
00:07:05,320 --> 00:07:12,310
And then we can run eight physical ports in a single etha channel between the distribution layer and

90
00:07:12,310 --> 00:07:13,040
access layer.

91
00:07:13,540 --> 00:07:19,180
And in that case, we don't need spanning tree because even though physically we have six switches,

92
00:07:19,480 --> 00:07:24,870
virtually, we only have two switches with one virtual cable between them.

93
00:07:25,480 --> 00:07:28,960
As always, the caveats and things to be aware of when doing this.

94
00:07:29,620 --> 00:07:36,280
But that's the ultimate vision of link aggregation with Chessie aggregation and switch stacking.

95
00:07:36,490 --> 00:07:42,550
You're aggregating your physical distribution switches into one virtual switch and you're stacking your

96
00:07:42,550 --> 00:07:45,640
access layer switches into one virtual switch.

97
00:07:46,090 --> 00:07:52,270
The simplifies the network because you don't have to worry about optimizing spending tree and optimizing

98
00:07:52,270 --> 00:07:53,050
HSP.

99
00:07:53,770 --> 00:07:57,910
There's no need for HSP because we have one virtual aggregated switch.

100
00:07:58,300 --> 00:08:01,000
We would still run spanning tree in case there are problems.

101
00:08:01,420 --> 00:08:06,580
But from a spending tree point of view, both of these ports are falling because there's only one logical

102
00:08:06,580 --> 00:08:08,740
cable between the two switches.

103
00:08:09,340 --> 00:08:13,960
It really simplifies the management and configuration of a campus network.

104
00:08:14,960 --> 00:08:22,070
So in summary, stacking technologies and Chessie aggregation technologies allow you to simplify the

105
00:08:22,070 --> 00:08:27,760
management and configuration, as well as the forwarding of traffic in an Ethernet network.