1
00:00:00,000 --> 00:00:03,458
[AUDIO LOGO]

2
00:00:03,458 --> 00:00:07,910

3
00:00:07,910 --> 00:00:12,260
Finally, let's talk
about anycast RP.

4
00:00:12,260 --> 00:00:16,320
Fundamentally, this is
a fairly simple concept.

5
00:00:16,320 --> 00:00:20,360
We're simply going to have
multiple routers with most

6
00:00:20,360 --> 00:00:25,580
likely loopback interfaces
sharing the same IP address,

7
00:00:25,580 --> 00:00:28,250
advertise that
loopback interface

8
00:00:28,250 --> 00:00:31,670
into our routing
protocol, and then we're

9
00:00:31,670 --> 00:00:35,330
going to use that
address as the RP.

10
00:00:35,330 --> 00:00:38,550
Now, where the
problem comes in is

11
00:00:38,550 --> 00:00:44,060
what happens if a first hop
router closest to the source,

12
00:00:44,060 --> 00:00:48,470
registers a source
with its closest RP,

13
00:00:48,470 --> 00:00:54,000
and then we have a client make
a request on a last hop router,

14
00:00:54,000 --> 00:00:59,030
and when that last hop
router joins the star, G tree

15
00:00:59,030 --> 00:01:04,430
and tries to get that source,
it ends up on a different RP

16
00:01:04,430 --> 00:01:09,390
because its closest instance
is not the same device.

17
00:01:09,390 --> 00:01:13,380
We need some way to
synchronize, or you

18
00:01:13,380 --> 00:01:17,160
could say cluster, the
source information that's

19
00:01:17,160 --> 00:01:21,780
available on the different
RPs, so that if a source gets

20
00:01:21,780 --> 00:01:26,670
registered on one RP, it shares
that information with all

21
00:01:26,670 --> 00:01:31,110
of the other RPs that are
in the same anycast group,

22
00:01:31,110 --> 00:01:35,320
meaning using that same
shared anycast address.

23
00:01:35,320 --> 00:01:38,100
So that's the fundamental
idea of what we're

24
00:01:38,100 --> 00:01:39,790
going to be talking about here.

25
00:01:39,790 --> 00:01:41,820
So let's jump in,
talk about some

26
00:01:41,820 --> 00:01:43,810
of the more detailed
information,

27
00:01:43,810 --> 00:01:47,340
and then we'll wrap up with
looking at this on a diagram.

28
00:01:47,340 --> 00:01:49,900

29
00:01:49,900 --> 00:01:54,750
So anycast isn't exactly
a way of finding the RP,

30
00:01:54,750 --> 00:01:58,020
but it's a way of
running the RP.

31
00:01:58,020 --> 00:02:00,630
So the way this
works, again, we're

32
00:02:00,630 --> 00:02:03,540
going to have
either two or more.

33
00:02:03,540 --> 00:02:06,150
There's really no
limitation on this.

34
00:02:06,150 --> 00:02:08,460
But we're going to
have these loopback

35
00:02:08,460 --> 00:02:15,810
interfaces on different RPs that
all have the same IP address.

36
00:02:15,810 --> 00:02:19,020
That's the whole
idea behind anycast.

37
00:02:19,020 --> 00:02:24,420
We then advertise this
address into the IGP,

38
00:02:24,420 --> 00:02:28,740
and the IGP whatever its
metric is, so again, maybe this

39
00:02:28,740 --> 00:02:33,680
is cost, maybe this is hop count
if you're using RIP, or maybe

40
00:02:33,680 --> 00:02:39,090
EIGRP's bandwidth and delay,
but whatever the IGP says,

41
00:02:39,090 --> 00:02:42,780
that's the closest instance
we're going to go to.

42
00:02:42,780 --> 00:02:46,290
Then we configure
this common IP address

43
00:02:46,290 --> 00:02:50,740
that we've configured on
all these devices as the RP

44
00:02:50,740 --> 00:02:52,130
address.

45
00:02:52,130 --> 00:02:54,800
Now, as we mentioned
during the introduction,

46
00:02:54,800 --> 00:02:58,330
the problem is going to be if
a source is registered with one

47
00:02:58,330 --> 00:03:01,210
RP and then the
client tries to get

48
00:03:01,210 --> 00:03:04,300
its information from
the last hop router

49
00:03:04,300 --> 00:03:08,500
and the last hop router tries
to get to a different RP.

50
00:03:08,500 --> 00:03:13,750
So the RPs are going to have
to have exactly the same source

51
00:03:13,750 --> 00:03:17,440
information, or what
we call SA, which

52
00:03:17,440 --> 00:03:21,140
is source active information.

53
00:03:21,140 --> 00:03:23,560
Now, there are two
ways to do this.

54
00:03:23,560 --> 00:03:27,760
On most Cisco IOS
platforms, the choice

55
00:03:27,760 --> 00:03:33,400
is going to be MSDP or Multicast
Source Discovery Protocol.

56
00:03:33,400 --> 00:03:36,820
However, on some
newer platforms,

57
00:03:36,820 --> 00:03:40,630
such as the Cisco
Nexus platform, which

58
00:03:40,630 --> 00:03:42,700
is more focused on
the data center,

59
00:03:42,700 --> 00:03:47,500
that allows us to actually
use PIM to advertise

60
00:03:47,500 --> 00:03:49,450
the anycast information.

61
00:03:49,450 --> 00:03:52,570
I only bring this up
because usually when

62
00:03:52,570 --> 00:03:54,850
a feature gets added
to one platform,

63
00:03:54,850 --> 00:03:57,610
it will start going
to other platforms.

64
00:03:57,610 --> 00:04:00,040
So even if you
don't use Nexus, we

65
00:04:00,040 --> 00:04:03,730
may see this in a
future version of IOS.

66
00:04:03,730 --> 00:04:07,210
So PIM has sort of been
updated or modified

67
00:04:07,210 --> 00:04:09,250
to carry this
information, we just

68
00:04:09,250 --> 00:04:12,860
have to wait for that to
catch up with other platforms.

69
00:04:12,860 --> 00:04:16,240
Now, this anycast
IP address can be

70
00:04:16,240 --> 00:04:19,870
learned by any of the
other means we've already

71
00:04:19,870 --> 00:04:20,769
talked about.

72
00:04:20,769 --> 00:04:23,140
It can be statically configured.

73
00:04:23,140 --> 00:04:27,700
It can, of course, be the
RP advertised into auto-RP.

74
00:04:27,700 --> 00:04:31,030
And it can be used in BSR.

75
00:04:31,030 --> 00:04:37,240
In fact, you could have multiple
anycast IP addresses being

76
00:04:37,240 --> 00:04:41,230
advertised into
BSR, and then BSR

77
00:04:41,230 --> 00:04:44,350
could use its
group-based load sharing

78
00:04:44,350 --> 00:04:46,570
to do even more load sharing.

79
00:04:46,570 --> 00:04:49,210
Remember that
anycast RP is going

80
00:04:49,210 --> 00:04:52,040
to be doing geographical
load sharing.

81
00:04:52,040 --> 00:04:57,070
So this is by itself going
to provide redundancy

82
00:04:57,070 --> 00:05:00,130
because, of course,
if one RP goes down,

83
00:05:00,130 --> 00:05:04,570
it would simply stop being
advertised into the IGP,

84
00:05:04,570 --> 00:05:07,480
and we would go to
another instance.

85
00:05:07,480 --> 00:05:10,730
Mobility is sort
of the same thing.

86
00:05:10,730 --> 00:05:13,480
We don't really care where
this IP address moves

87
00:05:13,480 --> 00:05:16,810
to because it's being
advertised in the IGP,

88
00:05:16,810 --> 00:05:21,250
and it might change which
is closest to which device,

89
00:05:21,250 --> 00:05:25,270
but overall, it still can
be mobile, no problem.

90
00:05:25,270 --> 00:05:30,050
And we're doing geographic
load sharing here, of course,

91
00:05:30,050 --> 00:05:33,940
because we're going to the
closest instance of the IP

92
00:05:33,940 --> 00:05:35,030
address.

93
00:05:35,030 --> 00:05:38,440
Well, if you take
BSR and you add that

94
00:05:38,440 --> 00:05:40,750
to the geographical
load sharing,

95
00:05:40,750 --> 00:05:44,890
you're going to get
group-based geographical load

96
00:05:44,890 --> 00:05:49,090
sharing, which is a pretty
robust option to put these two

97
00:05:49,090 --> 00:05:51,140
types of load sharing together.

98
00:05:51,140 --> 00:05:56,680
So running BSR with
anycast RP is actually

99
00:05:56,680 --> 00:06:00,080
a pretty good choice in
a lot of environments.

100
00:06:00,080 --> 00:06:03,400
So let's bring up a diagram
here and just discuss

101
00:06:03,400 --> 00:06:05,650
a little bit more
about how this works

102
00:06:05,650 --> 00:06:10,750
and where we would need
MSDP or PIM to synchronize

103
00:06:10,750 --> 00:06:12,380
this information.

104
00:06:12,380 --> 00:06:16,270
So looking at this diagram,
we have loopback interfaces

105
00:06:16,270 --> 00:06:22,240
on router 1 and on router 2
that are the same address.

106
00:06:22,240 --> 00:06:25,660
We are configuring all
the devices on the network

107
00:06:25,660 --> 00:06:28,990
to use that address as the RP.

108
00:06:28,990 --> 00:06:32,530
But here's where the
problem of comes into play.

109
00:06:32,530 --> 00:06:35,680
Let's say this source comes
online and begins sending

110
00:06:35,680 --> 00:06:37,750
traffic onto the network.

111
00:06:37,750 --> 00:06:41,350
Well, router 1,
being an RP itself,

112
00:06:41,350 --> 00:06:46,150
would register this source
on this router, which

113
00:06:46,150 --> 00:06:48,550
means it would
immediately create

114
00:06:48,550 --> 00:06:53,980
an S, G entry on this
device, and that would be it.

115
00:06:53,980 --> 00:06:56,940
It has no clients, it
has nothing on its oil.

116
00:06:56,940 --> 00:07:00,820
This traffic would go nowhere
because as of right now,

117
00:07:00,820 --> 00:07:02,740
we don't have any clients.

118
00:07:02,740 --> 00:07:05,620
Now, in order to sort of
demonstrate our problem here,

119
00:07:05,620 --> 00:07:09,360
we're going to add some
costs to these links.

120
00:07:09,360 --> 00:07:11,430
We're going to say
that these cross links

121
00:07:11,430 --> 00:07:15,450
between these devices
have a cost of 100.

122
00:07:15,450 --> 00:07:17,790
All of the other
links are simply

123
00:07:17,790 --> 00:07:21,940
going to have a cost of
10, just to keep it simple.

124
00:07:21,940 --> 00:07:25,950
So basically, what would happen
if the client on the left

125
00:07:25,950 --> 00:07:28,680
were to join this
group on router 3,

126
00:07:28,680 --> 00:07:34,950
router 3 would follow the IGP
and send a join to router 1

127
00:07:34,950 --> 00:07:36,660
for the shared tree.

128
00:07:36,660 --> 00:07:39,780
Router 1 has the
S,G information,

129
00:07:39,780 --> 00:07:43,710
and this would proceed
with sparse mode as normal.

130
00:07:43,710 --> 00:07:45,820
This would not be a problem.

131
00:07:45,820 --> 00:07:48,090
Here's where the
problem comes in.

132
00:07:48,090 --> 00:07:53,830
Let's say this client on
router 4 joins the group.

133
00:07:53,830 --> 00:07:59,920
Router 4 would, of course,
send this request to router 2.

134
00:07:59,920 --> 00:08:05,980
It would try to open a star, G
to router 2 or a shared tree.

135
00:08:05,980 --> 00:08:10,540
Now, router 2 would do that
because it's got a client,

136
00:08:10,540 --> 00:08:15,100
but all it would
have is a star, G

137
00:08:15,100 --> 00:08:20,620
entry because router 2 doesn't
know where the source is.

138
00:08:20,620 --> 00:08:25,870
What we need is that method of
sending this information over

139
00:08:25,870 --> 00:08:27,430
to this RP.

140
00:08:27,430 --> 00:08:30,580
This is that SA message
we were talking about,

141
00:08:30,580 --> 00:08:33,400
or the source active
message, and this

142
00:08:33,400 --> 00:08:37,600
is what's going to be
exchanged with either MSDP

143
00:08:37,600 --> 00:08:42,700
or on some platforms,
again, Nexus, PIM.

144
00:08:42,700 --> 00:08:45,880
But this is why we need
this additional protocol

145
00:08:45,880 --> 00:08:50,440
to keep the rendezvous
points in sync so they all

146
00:08:50,440 --> 00:08:55,380
know about the sources that
are available on our network.

147
00:08:55,380 --> 00:09:00,740
So in this video, we took a
look at anycast RP, which again,

148
00:09:00,740 --> 00:09:03,590
is not really a way
to discover the IP,

149
00:09:03,590 --> 00:09:06,500
it's more of a
way to run the IP.

150
00:09:06,500 --> 00:09:08,870
And as we can see on
this diagram here,

151
00:09:08,870 --> 00:09:12,140
the overall concept is
we have multiple devices

152
00:09:12,140 --> 00:09:14,390
with the same IP address.

153
00:09:14,390 --> 00:09:17,480
That IP address is
being advertised

154
00:09:17,480 --> 00:09:22,640
either statically, or through
BSR, or through auto-RP.

155
00:09:22,640 --> 00:09:24,470
It really doesn't
matter what method

156
00:09:24,470 --> 00:09:28,610
we use to get this address out
there to all the other routers,

157
00:09:28,610 --> 00:09:32,700
but that address is
configured as the RP.

158
00:09:32,700 --> 00:09:36,590
Now, all the devices can
go to the closest instance

159
00:09:36,590 --> 00:09:38,480
of that address.

160
00:09:38,480 --> 00:09:41,120
That's the geographical
load sharing.

161
00:09:41,120 --> 00:09:46,670
But as we discussed here, the
problem is all of the RPs need

162
00:09:46,670 --> 00:09:49,430
to know about all
of the sources then.

163
00:09:49,430 --> 00:09:53,420
And we can do that using
either MSDP, or again,

164
00:09:53,420 --> 00:09:54,710
on some platforms.

165
00:09:54,710 --> 00:09:58,470
And may be coming to an
IOS-based router near you,

166
00:09:58,470 --> 00:10:02,980
we have a way to do it through
PIM on some of those platforms.

167
00:10:02,980 --> 00:10:05,430
So that wraps up
everything that we

168
00:10:05,430 --> 00:10:08,010
need to discuss in this skill.

169
00:10:08,010 --> 00:10:10,770
That leaves us with
nothing else to do,

170
00:10:10,770 --> 00:10:13,980
except to review in
the form of a quiz.

171
00:10:13,980 --> 00:10:17,140
And that's exactly what
we're going to do next.

172
00:10:17,140 --> 00:10:20,790
Until then, I hope this has
been informative for you,

173
00:10:20,790 --> 00:10:23,930
and I'd like to thank
you for viewing.

174
00:10:23,930 --> 00:10:27,000