1
00:00:00,000 --> 00:00:02,385
[AUDIO LOGO]

2
00:00:02,385 --> 00:00:07,640

3
00:00:07,640 --> 00:00:11,150
So during our introduction,
we mentioned the fact

4
00:00:11,150 --> 00:00:14,450
that segment routing completely
changes the way the control

5
00:00:14,450 --> 00:00:17,480
plane works in regard to MPLS.

6
00:00:17,480 --> 00:00:20,120
Specifically, how
the labels are going

7
00:00:20,120 --> 00:00:22,580
to be exchanged
between the devices.

8
00:00:22,580 --> 00:00:24,980
We're going to be
getting rid of LDP

9
00:00:24,980 --> 00:00:29,000
and, actually, RSVP as
well, and instead use

10
00:00:29,000 --> 00:00:33,200
our IGPs, OSPF, and IS-IS.

11
00:00:33,200 --> 00:00:35,840
We'll also take a
look in this video

12
00:00:35,840 --> 00:00:38,430
about some other options
we have available,

13
00:00:38,430 --> 00:00:42,470
such as using a controller
or even using BGP.

14
00:00:42,470 --> 00:00:44,840
So let's go ahead
and take a look

15
00:00:44,840 --> 00:00:48,230
at how we change the
control plane to handle

16
00:00:48,230 --> 00:00:51,220
segment routing.

17
00:00:51,220 --> 00:00:54,810
So one of the big advantages
to the changing of the control

18
00:00:54,810 --> 00:00:58,560
plane is, since we're not
using other protocols,

19
00:00:58,560 --> 00:01:03,270
we're just using our IGP
and/or other routing protocols

20
00:01:03,270 --> 00:01:04,860
that we're already using.

21
00:01:04,860 --> 00:01:09,430
This control plane for segment
routing is far less complex.

22
00:01:09,430 --> 00:01:13,130

23
00:01:13,130 --> 00:01:15,560
We're using those
routing protocols

24
00:01:15,560 --> 00:01:20,420
that were already in place
to distribute the labels.

25
00:01:20,420 --> 00:01:22,560
And since that's
what we're doing,

26
00:01:22,560 --> 00:01:25,490
there's no dedicated protocol.

27
00:01:25,490 --> 00:01:30,680
So that means that LDP and
RSVP are no longer going

28
00:01:30,680 --> 00:01:33,060
to be used in the environment.

29
00:01:33,060 --> 00:01:34,550
So, of course,
instead, that means

30
00:01:34,550 --> 00:01:37,760
we're going to be using
either our IGP, which

31
00:01:37,760 --> 00:01:40,460
in the case of IS-IS
means we're going

32
00:01:40,460 --> 00:01:44,060
to be using new TLVs,
or type-length-values,

33
00:01:44,060 --> 00:01:48,170
and in the case of OSPF, it's
going to use something called

34
00:01:48,170 --> 00:01:50,600
opaque LSAs.

35
00:01:50,600 --> 00:01:54,440
Specifically, it
uses LSA type 10,

36
00:01:54,440 --> 00:01:58,800
and we'll talk about this in
more detail a little bit later.

37
00:01:58,800 --> 00:02:01,790
Now, as we mentioned,
we can also use BGP.

38
00:02:01,790 --> 00:02:06,550
Specifically, we use something
called BGP-LU, or labeled

39
00:02:06,550 --> 00:02:07,050
unicast.

40
00:02:07,050 --> 00:02:09,940

41
00:02:09,940 --> 00:02:14,080
Another option is to
use an SDN controller.

42
00:02:14,080 --> 00:02:17,650
SDN, if you're not familiar,
is software-defined networking.

43
00:02:17,650 --> 00:02:21,970
And what this does most
commonly is facilitates

44
00:02:21,970 --> 00:02:25,990
that multidomain SRTE,
which is the segment routing

45
00:02:25,990 --> 00:02:27,460
traffic engineering.

46
00:02:27,460 --> 00:02:29,530
For that multidomain
function, it's

47
00:02:29,530 --> 00:02:32,890
going to use BGP-LS,
or link state,

48
00:02:32,890 --> 00:02:38,530
and it uses something called
PCEP, or path computation

49
00:02:38,530 --> 00:02:41,380
element protocol.

50
00:02:41,380 --> 00:02:45,910
So as an example, we could
have a situation here where

51
00:02:45,910 --> 00:02:49,780
we have a controller, which
can be something as simple

52
00:02:49,780 --> 00:02:52,000
as an iOS XR router.

53
00:02:52,000 --> 00:02:57,370
Then the other devices
connect to this using PCEP.

54
00:02:57,370 --> 00:03:01,510
So these connections would
be using that protocol.

55
00:03:01,510 --> 00:03:04,960
Now, the individual
devices have names as well.

56
00:03:04,960 --> 00:03:06,850
The controller is
actually called

57
00:03:06,850 --> 00:03:11,470
the PCE, which is the
path computation element,

58
00:03:11,470 --> 00:03:17,200
and the clients are PCCs,
which are your path computation

59
00:03:17,200 --> 00:03:18,380
clients.

60
00:03:18,380 --> 00:03:21,580
So this is how this
entire SDN system works.

61
00:03:21,580 --> 00:03:24,430
And, again, keep in
mind that this is mostly

62
00:03:24,430 --> 00:03:26,950
used for traffic engineering.

63
00:03:26,950 --> 00:03:30,550
We'll be taking a much deeper
look at all of these elements

64
00:03:30,550 --> 00:03:34,270
and how this whole system
works when we get to our skill

65
00:03:34,270 --> 00:03:37,540
on segment routing
traffic engineering.

66
00:03:37,540 --> 00:03:39,270
So let's focus in
a little bit more

67
00:03:39,270 --> 00:03:42,220
on the changes we need to
do our routing protocols.

68
00:03:42,220 --> 00:03:49,080
So for IS-IS, it is going to
support both IPv4 and IPv6.

69
00:03:49,080 --> 00:03:54,300
As we mentioned, it uses new
TLVs, or type-length-values,

70
00:03:54,300 --> 00:03:57,240
for that SID advertisement.

71
00:03:57,240 --> 00:04:02,250
Specifically, it can handle
prefix SIDS, adjacency SIDS,

72
00:04:02,250 --> 00:04:05,730
and prefix-to-SID mappings.

73
00:04:05,730 --> 00:04:10,320
This is fully supported with
both level 1 and level 2 IS-IS

74
00:04:10,320 --> 00:04:14,250
routing, and it can
handle PHP, which is

75
00:04:14,250 --> 00:04:16,050
your penultimate hop popping.

76
00:04:16,050 --> 00:04:19,240
Or it can also
handle explicit null,

77
00:04:19,240 --> 00:04:22,740
which is, of course,
if you disable PHP.

78
00:04:22,740 --> 00:04:25,380
But IS-IS can
fully support both,

79
00:04:25,380 --> 00:04:29,730
whether that's enabled
on the MPLS stack or not.

80
00:04:29,730 --> 00:04:33,850
OSPF is going to be modified
in a very similar way.

81
00:04:33,850 --> 00:04:36,620
However, it does have
a few differences.

82
00:04:36,620 --> 00:04:41,920
First, it only supports
IPv4, at least right now.

83
00:04:41,920 --> 00:04:44,020
Of course, things will change.

84
00:04:44,020 --> 00:04:47,710
I'm sure that OSPF version
3 will be supported

85
00:04:47,710 --> 00:04:51,230
at some point, but at the
time of this recording,

86
00:04:51,230 --> 00:04:53,120
it's currently not.

87
00:04:53,120 --> 00:04:56,950
Now, this does, of course,
use OSPF extensions.

88
00:04:56,950 --> 00:05:02,020
We mentioned that it uses opaque
LSAs, specifically type 10,

89
00:05:02,020 --> 00:05:04,610
and then it can be
used for your prefix

90
00:05:04,610 --> 00:05:08,020
SIDs and your adjacency SIDs.

91
00:05:08,020 --> 00:05:11,500
These, by the way, are
carried in different subtypes

92
00:05:11,500 --> 00:05:13,270
of the type 10.

93
00:05:13,270 --> 00:05:15,910
We'll see that in
a lot more detail

94
00:05:15,910 --> 00:05:18,530
when we get to our
implementation skill.

95
00:05:18,530 --> 00:05:21,070
We see this on the command
line, and we actually

96
00:05:21,070 --> 00:05:23,290
dig into the OSPF
database a bit.

97
00:05:23,290 --> 00:05:27,220
So they're all LSA type
10s, but then there's

98
00:05:27,220 --> 00:05:31,630
subtypes under that that carry
the different types of SIDs

99
00:05:31,630 --> 00:05:34,160
that we need here
for segment routing.

100
00:05:34,160 --> 00:05:37,150
Now, sort of along the same
lines as supporting level 1

101
00:05:37,150 --> 00:05:40,810
level 2 in IS-IS,
this also fully

102
00:05:40,810 --> 00:05:44,980
supports multiarea from
an OSPF perspective,

103
00:05:44,980 --> 00:05:50,450
and it also supports
PHP and explicit null.

104
00:05:50,450 --> 00:05:52,550
So those things are the same.

105
00:05:52,550 --> 00:05:55,570
And, as we mentioned,
BGP can also

106
00:05:55,570 --> 00:05:58,090
be used to support
segment routing.

107
00:05:58,090 --> 00:06:01,570
We use a BGP prefix SID.

108
00:06:01,570 --> 00:06:06,070
This is generally used
between BGP domains

109
00:06:06,070 --> 00:06:09,700
to establish a full
label switch path.

110
00:06:09,700 --> 00:06:11,680
Remember, that
wouldn't normally be

111
00:06:11,680 --> 00:06:15,400
possible outside
of a single IGP.

112
00:06:15,400 --> 00:06:18,100
That would break the
label switch path.

113
00:06:18,100 --> 00:06:22,570
The segment is then
attached to the BGP prefix,

114
00:06:22,570 --> 00:06:24,520
and this, of course,
will have to be

115
00:06:24,520 --> 00:06:29,380
global within the entire
SR domain, meaning

116
00:06:29,380 --> 00:06:31,990
not just within,
of course, an IGP,

117
00:06:31,990 --> 00:06:35,110
because now we're joining
multiple IGPs together,

118
00:06:35,110 --> 00:06:38,570
or multiple IGP
domains together.

119
00:06:38,570 --> 00:06:43,280
So this is going to have to
be unique across everything.

120
00:06:43,280 --> 00:06:47,740
We also have a feature where
we support peering segments

121
00:06:47,740 --> 00:06:52,570
for egress peer
engineering, or EPE.

122
00:06:52,570 --> 00:06:59,750
This is used within a BGP domain
to control outbound traffic.

123
00:06:59,750 --> 00:07:02,260
So this is to choose
which exit you

124
00:07:02,260 --> 00:07:06,650
wish to use when going into
another BGP environment.

125
00:07:06,650 --> 00:07:09,410
In this category, there
are three families.

126
00:07:09,410 --> 00:07:13,300
We have the peer node segment,
the peer adjacency segment,

127
00:07:13,300 --> 00:07:16,250
and the peer set segment.

128
00:07:16,250 --> 00:07:19,330
Now, keep in mind, those
are all peer segments.

129
00:07:19,330 --> 00:07:22,810
They're just different
types of peer segments.

130
00:07:22,810 --> 00:07:25,300
That peer segment
is then going to be

131
00:07:25,300 --> 00:07:29,620
used to steer that traffic
over specific links

132
00:07:29,620 --> 00:07:31,810
to another BGP peer.

133
00:07:31,810 --> 00:07:34,120
That's why it's a peer segment.

134
00:07:34,120 --> 00:07:38,530
As of right now, there are
currently three peering segment

135
00:07:38,530 --> 00:07:40,310
identifiers defined.

136
00:07:40,310 --> 00:07:44,320
We have the peer node
SID, which is used

137
00:07:44,320 --> 00:07:46,690
to forward on any interface.

138
00:07:46,690 --> 00:07:50,050
We have the peer
adjacency SID, which

139
00:07:50,050 --> 00:07:53,590
is only used to forward
on the related interface.

140
00:07:53,590 --> 00:07:56,800
And then we have
the peer set SID,

141
00:07:56,800 --> 00:08:03,700
which is used to forward on any
interface to a set of peers.

142
00:08:03,700 --> 00:08:08,160
So in this video, we took a
closer look at the changes

143
00:08:08,160 --> 00:08:10,920
to the control plane
for segment routing.

144
00:08:10,920 --> 00:08:16,710
Specifically, we talked about
how IS-IS and OSPF have both

145
00:08:16,710 --> 00:08:20,580
been modified or updated
to support carrying

146
00:08:20,580 --> 00:08:22,860
labels for segment routing.

147
00:08:22,860 --> 00:08:26,400
We also talked about how we
could use an SDN controller

148
00:08:26,400 --> 00:08:28,740
for this, which, by the
way, can be something

149
00:08:28,740 --> 00:08:31,650
as simple as running
it on iOS XR,

150
00:08:31,650 --> 00:08:34,750
and we will be doing
that in a future skill.

151
00:08:34,750 --> 00:08:36,600
But the next thing
we need to look at

152
00:08:36,600 --> 00:08:38,710
are the different segment types.

153
00:08:38,710 --> 00:08:41,370
We've been talking a
whole lot about SIDs,

154
00:08:41,370 --> 00:08:44,770
but it's time to focus in on
some of those different types.

155
00:08:44,770 --> 00:08:47,770
And that's exactly what we're
going to be covering next.

156
00:08:47,770 --> 00:08:51,120
Until then, I hope this has
been informative for you,

157
00:08:51,120 --> 00:08:54,260
and I'd like to thank
you for viewing.

158
00:08:54,260 --> 00:08:58,000