WEBVTT

1
00:00.000 --> 00:07.760
So the claim at this point is that we have a set of win challenges and that SD-

2
00:07.760 --> 00:08.080
WAN is

3
00:08.080 --> 00:10.840
going to solve some of these problems for us.

4
00:10.840 --> 00:13.540
In order to prove that to ourselves, let's go ahead and list a few of the

5
00:13.540 --> 00:14.160
biggest win

6
00:14.160 --> 00:18.130
challenges we have and then talk about how this fancy new solution of ours is

7
00:18.130 --> 00:18.760
going to

8
00:18.760 --> 00:21.000
address the challenges.

9
00:21.000 --> 00:25.960
So first of all, as we talked about, we have routing complexities.

10
00:25.960 --> 00:29.320
The routing complexities are pretty enormous, really.

11
00:29.320 --> 00:34.860
I mean, for large organizations that are dealing with thousands of sites,
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

11
00:29.320 --> 00:34.860
I mean, for large organizations that are dealing with thousands of sites,

12
00:34.860 --> 00:35.340
trying to wrap

13
00:35.340 --> 00:38.560
our arms around all of that, but we've talked about some specifics.

14
00:38.560 --> 00:46.120
We have to share routes, for example, with the SP when it comes to MPLS.

15
00:46.120 --> 00:50.290
And that makes it so we have to do routing redistribution and prevent routing

16
00:50.290 --> 00:51.040
loops and

17
00:51.040 --> 00:52.720
all of those types of things.

18
00:52.720 --> 00:58.020
We also talked about application-specific direction, meaning that maybe I want

19
00:58.020 --> 00:58.760
voiceover

20
00:58.760 --> 01:03.620
IP to go out MPLS and I want the rest of my traffic to go out the internet and
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

20
00:58.760 --> 01:03.620
IP to go out MPLS and I want the rest of my traffic to go out the internet and

21
01:03.620 --> 01:04.160
I want

22
01:04.160 --> 01:09.720
to make sure that different applications maybe are using one or the other.

23
01:09.720 --> 01:13.710
So just the idea of directing out a particular win link depending on the

24
01:13.710 --> 01:14.840
traffic type is

25
01:14.840 --> 01:18.340
something that introduces a massive amount of complexities to traditional

26
01:18.340 --> 01:19.080
networking.

27
01:19.080 --> 01:24.560
Lastly, some amount of failover can be challenging.

28
01:24.560 --> 01:28.890
And again, it's not so much if a circuit goes down because routing protocols

29
01:28.890 --> 01:29.640
largely take

30
01:29.640 --> 01:31.080
care of that for us.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

30
01:29.640 --> 01:31.080
care of that for us.

31
01:31.080 --> 01:38.960
But what about not just outages, but also as we said, degradation.

32
01:38.960 --> 01:44.160
So if a particular circuit's having problems on this day for an hour, can we

33
01:44.160 --> 01:45.160
automatically

34
01:45.160 --> 01:50.250
divert traffic out the other win link and then allow us to detect when the

35
01:50.250 --> 01:51.160
first one is

36
01:51.160 --> 01:52.600
good and swing back over.

37
01:52.600 --> 01:54.600
So failover and failback.

38
01:54.600 --> 01:56.600
So lots of routing complexities.

39
01:56.600 --> 01:59.800
Now on top of all of this, we've got a lot of win circuits.

40
01:59.800 --> 02:02.280
We have a lot of devices that are out there.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

40
01:59.800 --> 02:02.280
We have a lot of devices that are out there.

41
02:02.280 --> 02:05.320
So just management of all these devices.

42
02:05.320 --> 02:09.150
When we're talking about software-defined networking one of the big benefits

43
02:09.150 --> 02:09.680
typically

44
02:09.680 --> 02:12.400
is in how we manage the system.

45
02:12.400 --> 02:15.680
Because we want to make sure that we can manage all these devices, but we don't

46
02:15.680 --> 02:16.600
have to

47
02:16.600 --> 02:21.000
want to have to log in and manage all these devices on top of being very

48
02:21.000 --> 02:22.280
specific in regards

49
02:22.280 --> 02:26.240
to our configurations of so many different policies and protocols.

50
02:26.240 --> 02:28.320
We also have a security problem.

51
02:28.320 --> 02:34.520
We've already mentioned that the internet has no encryption by default.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

51
02:28.320 --> 02:34.520
We've already mentioned that the internet has no encryption by default.

52
02:34.520 --> 02:38.600
And so we might want to encrypt our traffic that's going across the internet.

53
02:38.600 --> 02:40.520
And most organizations do that.

54
02:40.520 --> 02:43.040
But this isn't just about the internet.

55
02:43.040 --> 02:45.080
It's also about MPLS circuits.

56
02:45.080 --> 02:49.430
Now we say, well, why would we need to encrypt our traffic across an MPLS

57
02:49.430 --> 02:50.280
circuit?

58
02:50.280 --> 02:51.280
It's more private.

59
02:51.280 --> 02:53.280
It's not the public internet.

60
02:53.280 --> 02:58.000
Well MPLS is a multi-tenant environment.

61
02:58.000 --> 03:01.760
Meaning that this is a service provider that has many customers.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

61
02:58.000 --> 03:01.760
Meaning that this is a service provider that has many customers.

62
03:01.760 --> 03:06.960
And they might accidentally send our traffic to the wrong location.

63
03:06.960 --> 03:11.320
Maybe directly to a location that's logging all of the information.

64
03:11.320 --> 03:15.490
And so we just shared secrets by accident through no fault of our own

65
03:15.490 --> 03:17.920
technically to a different

66
03:17.920 --> 03:18.920
organization.

67
03:18.920 --> 03:23.330
That matter if there is a rogue employee working for the service provider that

68
03:23.330 --> 03:23.960
just wants to

69
03:23.960 --> 03:24.960
capture our traffic.

70
03:24.960 --> 03:30.160
I mean, we're opening ourselves up to risk that our traffic could become
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

70
03:24.960 --> 03:30.160
I mean, we're opening ourselves up to risk that our traffic could become

71
03:30.160 --> 03:31.480
available in some

72
03:31.480 --> 03:32.480
way.

73
03:32.480 --> 03:36.690
And so we have to worry about the fact that we have no encryption even over

74
03:36.690 --> 03:37.560
circuits such

75
03:37.560 --> 03:39.720
as MPLS.

76
03:39.720 --> 03:46.280
And then lastly, we mentioned it before, but the expense of our WANs.

77
03:46.280 --> 03:50.000
So we're especially talking about MPLS circuits.

78
03:50.000 --> 03:53.760
We're talking about maybe T3s and E3s.

79
03:53.760 --> 03:56.640
To some extent, T1s and E1s as well.

80
03:56.640 --> 04:02.300
If we have dark fiber services or for that matter, lit fiber services, a lot of
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

80
03:56.640 --> 04:02.300
If we have dark fiber services or for that matter, lit fiber services, a lot of

81
04:02.300 --> 04:03.480
these WAN

82
04:03.480 --> 04:06.720
types can be very expensive.

83
04:06.720 --> 04:13.120
And so we find that we need to have one of them, maybe it's MPLS, as a reliable

84
04:13.120 --> 04:13.880
option.

85
04:13.880 --> 04:18.240
And then we try to have our redundant circuit via the internet because it's

86
04:18.240 --> 04:19.360
less expensive.

87
04:19.360 --> 04:24.160
But we can't replace both of these with the internet and be very inexpensive

88
04:24.160 --> 04:25.080
overall because

89
04:25.080 --> 04:28.680
of all of the downfalls that the internet brings with it.

90
04:28.680 --> 04:31.640
So that is our list of challenges that we need to address.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

90
04:28.680 --> 04:31.640
So that is our list of challenges that we need to address.

91
04:31.640 --> 04:37.540
So how does software defined WAN come into play as we start to unpack the

92
04:37.540 --> 04:39.000
architecture,

93
04:39.000 --> 04:43.120
which we really still have yet to do, but at a high level, what are we trying

94
04:43.120 --> 04:44.000
to accomplish

95
04:44.000 --> 04:52.000
with SD WAN and the solutions that it claims to bring?

96
04:52.000 --> 04:54.160
Well let's just go through these one by one.

97
04:54.160 --> 04:57.440
And we're going to start with the routing complexities.

98
04:57.440 --> 05:02.140
So the routing complexities are going to be resolved because we're adopting a
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

98
04:57.440 --> 05:02.140
So the routing complexities are going to be resolved because we're adopting a

99
05:02.140 --> 05:02.800
software

100
05:02.800 --> 05:06.800
defined architecture or a software defined network.

101
05:06.800 --> 05:11.580
And I think back to when we studied VXLAN, and we talked about how we've got

102
05:11.580 --> 05:12.440
different

103
05:12.440 --> 05:17.770
architectures and we can leverage VXLAN over different fabrics, which allows us

104
05:17.770 --> 05:18.480
to pretty

105
05:18.480 --> 05:24.400
much create the network that I want.

106
05:24.400 --> 05:28.080
And that's compared to the network that I have.

107
05:28.080 --> 05:32.720
So what I have is a bunch of circuits.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

107
05:28.080 --> 05:32.720
So what I have is a bunch of circuits.

108
05:32.720 --> 05:33.960
But what do I want?

109
05:33.960 --> 05:38.250
I might want a completely different network and I'm going to use tunnels in

110
05:38.250 --> 05:39.640
order to create

111
05:39.640 --> 05:40.640
that network.

112
05:40.640 --> 05:43.080
And tunnels are like the Swiss Army knife of networking.

113
05:43.080 --> 05:47.890
I can kind of do whatever I want with tunnels because I just direct traffic and

114
05:47.890 --> 05:48.800
encapsulate

115
05:48.800 --> 05:50.640
it from one router to another.

116
05:50.640 --> 05:55.860
I might have this crazy network with a bunch of routers and I have to make sure

117
05:55.860 --> 05:56.560
that every

118
05:56.560 --> 06:00.480
router knows where the destination is because when a packet arrives there, they
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

118
05:56.560 --> 06:00.480
router knows where the destination is because when a packet arrives there, they

119
06:00.480 --> 06:00.920
have to

120
06:00.920 --> 06:04.200
forward it towards that destination.

121
06:04.200 --> 06:10.880
Or I can say, you know what, I'm going to build a tunnel from one router to the

122
06:10.880 --> 06:11.040
other and

123
06:11.040 --> 06:14.920
then all of these routers in the middle just have to know how to get to that

124
06:14.920 --> 06:15.880
router because

125
06:15.880 --> 06:18.040
I'm just sending it essentially to that router.

126
06:18.040 --> 06:19.840
I mean, encapsulating it here.

127
06:19.840 --> 06:23.910
I'm forwarding it to this router and so everybody's forwarding it to a

128
06:23.910 --> 06:25.640
different destination,

129
06:25.640 --> 06:30.730
not to the final destination, but to the router that is terminating the tunnel,
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

129
06:25.640 --> 06:30.730
not to the final destination, but to the router that is terminating the tunnel,

130
06:30.730 --> 06:31.360
which is

131
06:31.360 --> 06:32.600
kind of interesting.

132
06:32.600 --> 06:37.120
So now we're hiding those inner headers and we're able to route between the

133
06:37.120 --> 06:38.240
edge routers.

134
06:38.240 --> 06:42.390
This is really important because now we don't need to worry about sharing

135
06:42.390 --> 06:43.320
routes with our

136
06:43.320 --> 06:44.840
service provider.

137
06:44.840 --> 06:48.000
The service provider's routers would be the ones here in the middle.

138
06:48.000 --> 06:51.450
If I'm forwarding traffic through an MPLS circuit, they have to know where the

139
06:51.450 --> 06:52.160
destination

140
06:52.160 --> 06:56.160
network is, which is why I usually have to share my routes with them.

141
06:56.160 --> 06:59.560
But if I'm building tunnels, well, I'm just going from one public IP address or

142
06:59.560 --> 06:59.960
at least

143
06:59.960 --> 07:05.280
an IP address that's in their network to an IP address that is also in their
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

143
06:59.960 --> 07:05.280
an IP address that's in their network to an IP address that is also in their

144
07:05.280 --> 07:06.280
network.

145
07:06.280 --> 07:08.400
I should need to share routes with them at all.

146
07:08.400 --> 07:12.190
They should know how to get from one of their subnets to another one of their

147
07:12.190 --> 07:12.840
subnets,

148
07:12.840 --> 07:16.480
which is absolutely amazing.

149
07:16.480 --> 07:21.680
Furthermore, because of tunneling architectures, I can now build tunnels

150
07:21.680 --> 07:23.280
through two different

151
07:23.280 --> 07:24.440
circuit types.

152
07:24.440 --> 07:30.930
So now I can have my router that is connected into whatever this circuit type
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

152
07:24.440 --> 07:30.930
So now I can have my router that is connected into whatever this circuit type

153
07:30.930 --> 07:31.560
is.

154
07:31.560 --> 07:33.320
We're just going to call it number one.

155
07:33.320 --> 07:37.080
And whatever this circuit type is down here, we're going to call it number two.

156
07:37.080 --> 07:40.720
And I've got circuits going again to this other router.

157
07:40.720 --> 07:43.960
Well, first of all, I am building the tunnel.

158
07:43.960 --> 07:46.560
I'm going to build a tunnel through this service provider.

159
07:46.560 --> 07:49.920
I'm going to build another tunnel through this service provider.

160
07:49.920 --> 07:55.870
And now I have a whole lot more control from an application perspective,

161
07:55.870 --> 07:56.640
because I can

162
07:56.640 --> 08:00.090
decide based on the application as defined in software, whether I'm going to
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

162
07:56.640 --> 08:00.090
decide based on the application as defined in software, whether I'm going to

163
08:00.090 --> 08:00.600
send it out

164
08:00.600 --> 08:03.040
this tunnel or that tunnel.

165
08:03.040 --> 08:07.930
And that's the end of the story from a route switch architecture perspective

166
08:07.930 --> 08:08.560
with an SD

167
08:08.560 --> 08:09.560
WAN.

168
08:09.560 --> 08:12.960
I don't have to worry about what's happening inside either of these underlays.

169
08:12.960 --> 08:17.200
I just have to worry about inserting a particular traffic type into a

170
08:17.200 --> 08:18.600
particular tunnel.

171
08:18.600 --> 08:22.200
And that's going to help me take care of things from that perspective.

172
08:22.200 --> 08:29.520
I can also run a lot of metrics and detect when we have degradation.

173
08:29.520 --> 08:33.220
And if I can control which application through software is going out which of
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

173
08:29.520 --> 08:33.220
And if I can control which application through software is going out which of

174
08:33.220 --> 08:34.200
these tunnels,

175
08:34.200 --> 08:39.070
what all I have to do is pass that information up to the application controller

176
08:39.070 --> 08:39.720
, which will

177
08:39.720 --> 08:42.120
determine which tunnel I'm going to use.

178
08:42.120 --> 08:48.780
So the routing complexities are going to be much reduced when it comes to SD W

179
08:48.780 --> 08:49.080
AN.

180
08:49.080 --> 08:53.860
And at a minimum, we can see very clearly how we no longer have to share routes

181
08:53.860 --> 08:54.320
with the

182
08:54.320 --> 08:56.040
service provider.

183
08:56.040 --> 08:59.120
Now the management is a relatively quick conversation.

184
08:59.120 --> 09:03.700
We are going to manage all this through a single pane of glass as we like to
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

184
08:59.120 --> 09:03.700
We are going to manage all this through a single pane of glass as we like to

185
09:03.700 --> 09:04.280
say, meaning

186
09:04.280 --> 09:11.610
that I have a single controller, an SDN controller that I log into and I

187
09:11.610 --> 09:13.880
perform all of these configurations.

188
09:13.880 --> 09:19.360
When we say, talking about app control and all these metrics and analyzing all

189
09:19.360 --> 09:20.120
of that,

190
09:20.120 --> 09:25.810
that is going to be the responsibility of the SDN control plane and management

191
09:25.810 --> 09:26.560
plane.

192
09:26.560 --> 09:30.120
And that's going to be seen here via the single pane of glass.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

192
09:26.560 --> 09:30.120
And that's going to be seen here via the single pane of glass.

193
09:30.120 --> 09:34.070
I as the network admin, I can log into the single pane of glass rather than

194
09:34.070 --> 09:34.920
having to log

195
09:34.920 --> 09:40.240
into lots and lots of different routers.

196
09:40.240 --> 09:44.400
Next up from a security perspective, we are going to not just use tunnels, but

197
09:44.400 --> 09:45.040
we are going

198
09:45.040 --> 09:49.200
to use encrypted tunnels.

199
09:49.200 --> 09:54.910
Specifically, the tunnels we're talking about here, these are going to use IP

200
09:54.910 --> 09:56.080
sec, allowing

201
09:56.080 --> 10:01.130
us to encrypt our tunnels regardless of whether this is the internet circuit
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

201
09:56.080 --> 10:01.130
us to encrypt our tunnels regardless of whether this is the internet circuit

202
10:01.130 --> 10:01.960
and the MPLS

203
10:01.960 --> 10:05.360
circuit down here or vice versa, or maybe they're both MPLS circuits.

204
10:05.360 --> 10:06.360
It doesn't matter.

205
10:06.360 --> 10:08.880
I'm going to encrypt it no matter what.

206
10:08.880 --> 10:13.660
So now, not only am I not sharing my routes with my service provider, I'm not

207
10:13.660 --> 10:14.720
sending my

208
10:14.720 --> 10:17.480
data in plain text to them either.

209
10:17.480 --> 10:21.840
So overall, a big win from a security perspective.

210
10:21.840 --> 10:25.550
And last but not least, the cost from an IT perspective, we don't always care

211
10:25.550 --> 10:26.040
about the

212
10:26.040 --> 10:27.040
cost of a solution.

213
10:27.040 --> 10:32.480
We just want the best solution from an administrative management perspective.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

213
10:27.040 --> 10:32.480
We just want the best solution from an administrative management perspective.

214
10:32.480 --> 10:34.960
We like inexpensive solutions.

215
10:34.960 --> 10:39.240
This is an interesting conversation now because what have we said about the

216
10:39.240 --> 10:40.120
internet?

217
10:40.120 --> 10:43.800
The internet is cheap, but it's potentially unreliable.

218
10:43.800 --> 10:49.560
Now, when I say it's unreliable, it's usually not unreliable minute to minute.

219
10:49.560 --> 10:53.280
It's usually more like we have some congestion on the internet.

220
10:53.280 --> 10:55.440
I can't control quality of service as well.

221
10:55.440 --> 10:59.240
And therefore, there's going to be network drops at various times of day.

222
10:59.240 --> 11:04.000
So this unreliable nature, as long as it's not rampant, means that I could
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

222
10:59.240 --> 11:04.000
So this unreliable nature, as long as it's not rampant, means that I could

223
11:04.000 --> 11:04.800
probably run

224
11:04.800 --> 11:09.400
on the internet circuit 95% of the time.

225
11:09.400 --> 11:14.560
But what happens now, given that I have this app level of control, and I have

226
11:14.560 --> 11:15.480
the ability

227
11:15.480 --> 11:21.800
to detect when we have degradation, can I just deploy two internet circuits?

228
11:21.800 --> 11:24.400
We'll look at this topology up here.

229
11:24.400 --> 11:29.140
What happens if we have an internet circuit for topology one or circuit one,

230
11:29.140 --> 11:29.880
but we also

231
11:29.880 --> 11:34.070
have another internet circuit, a completely different service provider as my
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

231
11:29.880 --> 11:34.070
have another internet circuit, a completely different service provider as my

232
11:34.070 --> 11:35.800
second path.

233
11:35.800 --> 11:41.450
And so if we say, well, this is 95% reliable, and this one's 95% reliable, as

234
11:41.450 --> 11:42.440
long as the odds

235
11:42.440 --> 11:47.060
are pretty low that they're both unreliable at the same time, we, a lot of

236
11:47.060 --> 11:48.040
cases, can just

237
11:48.040 --> 11:50.040
adopt two different internet circuits.

238
11:50.040 --> 11:55.360
It at least allows us to have the conversation, and it truly is a conversation.

239
11:55.360 --> 12:00.520
This is not a recommendation from Cisco, or for me, for that matter, that we go
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

239
11:55.360 --> 12:00.520
This is not a recommendation from Cisco, or for me, for that matter, that we go

240
12:00.520 --> 12:01.000
out and

241
12:01.000 --> 12:02.400
cancel MPLS circuits.

242
12:02.400 --> 12:06.520
But we can at least have the conversation that, hey, if we live in a part of

243
12:06.520 --> 12:07.560
the world,

244
12:07.560 --> 12:12.540
and where we just got very reliable internet, then maybe we could have two

245
12:12.540 --> 12:13.880
internet circuits,

246
12:13.880 --> 12:16.800
and that would greatly drive the cost down.

247
12:16.800 --> 12:21.400
So that's going to be a potential that comes out of the SD-WAN world, which

248
12:21.400 --> 12:22.160
allows us to

249
12:22.160 --> 12:27.320
at least think outside the box of having to have MPLS circuits.

250
12:27.320 --> 12:32.040
So at this point, again, from a very high level, we can see that SD-WAN at
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

250
12:27.320 --> 12:32.040
So at this point, again, from a very high level, we can see that SD-WAN at

251
12:32.040 --> 12:32.880
least claims to

252
12:32.880 --> 12:36.840
have solutions to all of the different challenges that we have in our

253
12:36.840 --> 12:38.680
traditional WAN architectures.
