WEBVTT

1
00:00.180 --> 00:02.370
Now, the actual edge transport node

2
00:02.370 --> 00:04.080
could be one of two flavors.

3
00:04.080 --> 00:07.320
We could deploy the edge transport node as a VM,

4
00:07.320 --> 00:09.600
that's the most easy and popular option.

5
00:09.600 --> 00:12.090
Or if we actually have a ton of throughput

6
00:12.090 --> 00:13.200
that needs to be routed,

7
00:13.200 --> 00:15.900
we may choose to use a dedicated piece of hardware

8
00:15.900 --> 00:17.640
for an edge transport node,

9
00:17.640 --> 00:19.650
and that's where we use a physical server.

10
00:19.650 --> 00:21.150
And that implementation is also known

11
00:21.150 --> 00:25.080
as implementing our edge transport node on bare metal.

12
00:25.080 --> 00:27.300
And for our demonstration of our implementation here,

13
00:27.300 --> 00:31.067
let's implement the edge transport node as a VM.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

13
00:27.300 --> 00:31.067
let's implement the edge transport node as a VM.

14
00:31.067 --> 00:34.470
So let me go ahead and remove the physical bare metal option

15
00:34.470 --> 00:36.570
as far as an edge transport node goes,

16
00:36.570 --> 00:38.370
and let's plan on using a VM.

17
00:38.370 --> 00:41.070
It's also a good idea to have some fault tolerance as well.

18
00:41.070 --> 00:42.810
So if we're implementing, for example,

19
00:42.810 --> 00:44.970
an active standby or active-active scenario

20
00:44.970 --> 00:46.830
for a specific gateway,

21
00:46.830 --> 00:48.990
we're gonna wanna have more than just one VM.

22
00:48.990 --> 00:50.880
So in that case, we could deploy two VMs,

23
00:50.880 --> 00:52.560
two edge transport nodes,

24
00:52.560 --> 00:54.270
and then we could place them together

25
00:54.270 --> 00:56.100
inside of an edge node cluster

26
00:56.100 --> 00:58.800
and that would be a more reliable pool to pull from.

27
00:58.800 --> 01:00.780
So that way if we lose one of the VMs,
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

27
00:58.800 --> 01:00.780
So that way if we lose one of the VMs,

28
01:00.780 --> 01:02.940
the other one can still provide the functions

29
01:02.940 --> 01:05.010
for the layer three gateway services.

30
01:05.010 --> 01:06.000
And again, for our purposes,

31
01:06.000 --> 01:08.730
we're gonna be implementing the edge as VMs.

32
01:08.730 --> 01:09.960
So here's what I propose we do.

33
01:09.960 --> 01:12.150
Using the NSX manager,

34
01:12.150 --> 01:15.660
let's deploy two edge node VMs,

35
01:15.660 --> 01:17.340
and VMware makes it super easy

36
01:17.340 --> 01:21.270
to do right from the gooey interface on the NSX Manager.

37
01:21.270 --> 01:22.230
So as far as the VMs,

38
01:22.230 --> 01:25.980
let's go ahead and call them Edge-1 and Edge-2.

39
01:25.980 --> 01:27.810
It'd also be a great idea to make sure

40
01:27.810 --> 01:31.380
that whatever names you use are resolvable in DNS,
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

40
01:27.810 --> 01:31.380
that whatever names you use are resolvable in DNS,

41
01:31.380 --> 01:33.480
and that goes for everything else here as well

42
01:33.480 --> 01:36.510
as far as your ESXi hosts and your vCenter server

43
01:36.510 --> 01:37.830
and your NSX Manager,

44
01:37.830 --> 01:41.760
we wanna make sure everything is resolvable via DNS.

45
01:41.760 --> 01:43.620
We'd also wanna make sure we have accurate time

46
01:43.620 --> 01:44.880
with network time protocol

47
01:44.880 --> 01:46.350
to make sure everything's in sync.

48
01:46.350 --> 01:50.040
So again, for the VMs, we'll call one Edge-1 and one Edge-2.

49
01:50.040 --> 01:53.880
For the domain, we'll put 'em in ogit.local

50
01:53.880 --> 01:56.040
for the purposes of DNS resolution.

51
01:56.040 --> 01:57.060
And for the management address,

52
01:57.060 --> 01:58.320
here's my management network right here,

53
01:58.320 --> 02:01.350
it's on the 192.168.1 network.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

53
01:58.320 --> 02:01.350
it's on the 192.168.1 network.

54
02:01.350 --> 02:04.260
So the vCenter server is at .185.

55
02:04.260 --> 02:06.360
On the management network, the NSX Manager,

56
02:06.360 --> 02:08.670
the virtual IP address is at .10.

57
02:08.670 --> 02:13.410
ESXi host A is on the management network at .181

58
02:13.410 --> 02:17.460
and ESXi-B is on the management network at .182.

59
02:17.460 --> 02:19.830
So for these two edge VMs that we're gonna deploy

60
02:19.830 --> 02:21.600
for the management network for Edge-1,

61
02:21.600 --> 02:26.100
let's use 192.168.1.191 with a 24-bit mask.

62
02:26.100 --> 02:27.270
And then for Edge-2,

63
02:27.270 --> 02:29.520
we'll give it the IP address on the management network

64
02:29.520 --> 02:32.250
of 192.168.1.192.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

64
02:29.520 --> 02:32.250
of 192.168.1.192.

65
02:32.250 --> 02:35.070
And I have a default gateway off that management network,

66
02:35.070 --> 02:37.350
and that's a little router down here, it's at .1.

67
02:37.350 --> 02:39.030
Now, the next question we're gonna answer together

68
02:39.030 --> 02:41.880
in our planning step here is where should we run these VMs?

69
02:41.880 --> 02:45.360
Now, we could run them pretty much on any ESXi host

70
02:45.360 --> 02:46.890
that has the bandwidth to support it.

71
02:46.890 --> 02:51.890
So I've got a management cluster with ESXi-C and D.

72
02:52.110 --> 02:55.470
I've also got two ESXi hosts that are enabled for NSX,

73
02:55.470 --> 02:58.800
meaning they're NSX host transport nodes.

74
02:58.800 --> 03:00.540
So if we needed to, although it's not required,
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

74
02:58.800 --> 03:00.540
So if we needed to, although it's not required,

75
03:00.540 --> 03:03.030
we could also go ahead and put these edge VMs

76
03:03.030 --> 03:07.350
and have the ESXi host A and B support those as well.

77
03:07.350 --> 03:08.700
And as I consider the load

78
03:08.700 --> 03:11.340
that we currently have on our four ESXi hosts,

79
03:11.340 --> 03:14.370
I currently have the NSX Manager right here.

80
03:14.370 --> 03:16.350
It's running on ESXi-C

81
03:16.350 --> 03:20.460
And the vCenter is currently running as a VM on ESXi-D.

82
03:20.460 --> 03:22.980
So I think just for capacity purposes,

83
03:22.980 --> 03:26.160
we should go ahead and run one of these edge VMs

84
03:26.160 --> 03:30.120
on the competing resource, known and loved as ESXi-A.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

84
03:26.160 --> 03:30.120
on the competing resource, known and loved as ESXi-A.

85
03:30.120 --> 03:34.050
And this VM will go ahead and run over here on ESXi-B.

86
03:34.050 --> 03:36.270
But I do wanna make it clear that these two hosts

87
03:36.270 --> 03:37.530
where we're running these VMs,

88
03:37.530 --> 03:41.670
they do not have to be NSX specific transport nodes.

89
03:41.670 --> 03:43.740
They can just be regular ESXi host

90
03:43.740 --> 03:45.450
providing the compute resources

91
03:45.450 --> 03:47.520
as long as we're giving the VMs access

92
03:47.520 --> 03:49.080
to the network connections they need.

93
03:49.080 --> 03:51.330
So currently, each of these ESXi hosts

94
03:51.330 --> 03:54.690
has a port group that leads out to my management network.

95
03:54.690 --> 03:56.400
So as we deploy the VMs,

96
03:56.400 --> 03:59.220
we'll specify which interface we're gonna be using

97
03:59.220 --> 04:00.930
and how to connect that to the port group.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

97
03:59.220 --> 04:00.930
and how to connect that to the port group.

98
04:00.930 --> 04:02.307
That leads up to the management network.

99
04:02.307 --> 04:04.290
And the next element we get to plan for

100
04:04.290 --> 04:07.080
is the tunnel endpoint address base,

101
04:07.080 --> 04:09.360
including what VLANs we're gonna use

102
04:09.360 --> 04:10.590
for that transport network

103
04:10.590 --> 04:14.220
and also which physical uplink on the ESXi hosts.

104
04:14.220 --> 04:16.260
So here's what I propose we do.

105
04:16.260 --> 04:19.320
Let's go ahead and use ESXi-A and ESXi-B

106
04:19.320 --> 04:22.740
as the homes for Edge-1 and Edge-2 respectively.

107
04:22.740 --> 04:25.080
And for the tunnel endpoint on these VMs,

108
04:25.080 --> 04:28.500
let's use the IP dressing space of 33, et cetera.

109
04:28.500 --> 04:31.560
And on Edge-2, we'll use the addressing space of 44.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

109
04:28.500 --> 04:31.560
And on Edge-2, we'll use the addressing space of 44.

110
04:31.560 --> 04:35.640
Now, we could put all the TEPs in the same exact subnet.

111
04:35.640 --> 04:39.510
In fact, now with the current flavor of NSX version 4.x.

112
04:39.510 --> 04:42.750
you can also go ahead and put the TEPs for the edge nodes

113
04:42.750 --> 04:45.240
and also the TEPs for the host transport nodes.

114
04:45.240 --> 04:48.360
We could put all of them in the same exact VLAN,

115
04:48.360 --> 04:49.890
assuming that VLAN and that subnet

116
04:49.890 --> 04:51.660
was available across the board.

117
04:51.660 --> 04:53.730
However, for the purpose of learning,

118
04:53.730 --> 04:56.130
if we create separate subnets

119
04:56.130 --> 04:58.290
and have very distinct IP addresses,

120
04:58.290 --> 05:00.690
that way as these tunnels are being built
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

120
04:58.290 --> 05:00.690
that way as these tunnels are being built

121
05:00.690 --> 05:03.690
between host transport nodes and edge transport nodes,

122
05:03.690 --> 05:05.520
it'll be very, very clear

123
05:05.520 --> 05:07.650
who they're coming from and where they're going to.

124
05:07.650 --> 05:11.640
So ESXi-A has a TEP of 11.11.11.11.

125
05:11.640 --> 05:14.700
And ESXi-B has a TEP of 22.22.22.22.

126
05:14.700 --> 05:17.880
And that way if we make Edge-1, 33.33.33.33,

127
05:17.880 --> 05:20.010
and Edge-2, a whole bunch of fours there,

128
05:20.010 --> 05:22.560
it'll be super easy to identify the tunnel source

129
05:22.560 --> 05:25.320
and destination and also identify which edge node

130
05:25.320 --> 05:27.900
or host transport node it happens to be.

131
05:27.900 --> 05:31.980
So for Edge-1, let's go ahead and use vmnic three,
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

131
05:27.900 --> 05:31.980
So for Edge-1, let's go ahead and use vmnic three,

132
05:31.980 --> 05:33.150
uplink number three,

133
05:33.150 --> 05:36.210
and allow it exclusive use by this Edge VM.

134
05:36.210 --> 05:38.580
And then on that link, let's also go ahead

135
05:38.580 --> 05:40.410
and use VLAN 103,

136
05:40.410 --> 05:44.190
that in my network corresponds to the 33.33 address base.

137
05:44.190 --> 05:45.360
And then for Edge-2,

138
05:45.360 --> 05:47.670
which is gonna be running over here on ESXi-B,

139
05:47.670 --> 05:49.530
let's allocate it Uplink #3

140
05:49.530 --> 05:51.810
over here on this host ESXi-B.

141
05:51.810 --> 05:54.660
And over here, let's go ahead and use VLAN 104,

142
05:54.660 --> 05:58.470
which corresponds to my 44.44.44.44 address space

143
05:58.470 --> 05:59.970
on the transport network.

144
05:59.970 --> 06:01.980
And as a result of our final configuration,
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

144
05:59.970 --> 06:01.980
And as a result of our final configuration,

145
06:01.980 --> 06:04.290
which is going to include in a future set of videos,

146
06:04.290 --> 06:07.470
setting up the actual layer three gateways themselves,

147
06:07.470 --> 06:10.320
we could have tunnels from the VTEP on Edge-2

148
06:10.320 --> 06:13.940
that go to the host transport node on ESXi-A.

149
06:13.940 --> 06:16.380
We could set up a tunnel from Edge-1 from this TEP,

150
06:16.380 --> 06:18.960
going over to the host transport node functionality.

151
06:18.960 --> 06:20.850
And the TEP here on ESXi-B

152
06:20.850 --> 06:23.160
or from Edge-1 down to ESXi-A

153
06:23.160 --> 06:25.410
or from Edge-2 down to ESXi-B

154
06:25.410 --> 06:28.050
And of course, we've already seen this in a previous skill,

155
06:28.050 --> 06:29.580
we have a tunnel that can be set up

156
06:29.580 --> 06:31.800
between ESXi-A and ESXi-B
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

156
06:29.580 --> 06:31.800
between ESXi-A and ESXi-B

157
06:31.800 --> 06:34.320
in the event they have VMs that need traffic moved

158
06:34.320 --> 06:35.760
that are on a common segment.

159
06:35.760 --> 06:37.320
In our example, in the previous skill,

160
06:37.320 --> 06:39.570
it was segment 10 and segment 20.

161
06:39.570 --> 06:41.010
And I also think I left off one here.

162
06:41.010 --> 06:43.050
We also possibly could have a tunnel segment

163
06:43.050 --> 06:46.590
between Edge-1 and Edge-2 because they're separate VMs.

164
06:46.590 --> 06:48.480
So in any of these NSX nodes,

165
06:48.480 --> 06:51.210
whether they're edge nodes or host transport nodes

166
06:51.210 --> 06:52.710
need to communicate with each other.

167
06:52.710 --> 06:54.537
They're gonna be using their respective uplinks

168
06:54.537 --> 06:56.340
and the respective VLANs

169
06:56.340 --> 06:57.780
for the transport network

170
06:57.780 --> 07:01.230
as they send packets back and forth over the tunnels.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

170
06:57.780 --> 07:01.230
as they send packets back and forth over the tunnels.

171
07:01.230 --> 07:02.063
And you might ask,

172
07:02.063 --> 07:04.620
"Well, how in the world is ESXi ever gonna need

173
07:04.620 --> 07:07.560
to send a tunnel over to Edge-1 or to Edge-2?"

174
07:07.560 --> 07:08.820
And here's a quick example of that.

175
07:08.820 --> 07:10.950
Let's imagine we have ESXi-A here

176
07:10.950 --> 07:13.020
and it has a little VM here,

177
07:13.020 --> 07:16.440
and that VM is connected to an NSX overlay segment.

178
07:16.440 --> 07:17.910
Let's call it segment 20.

179
07:17.910 --> 07:20.773
And the addressing space there is 20.0.0,

180
07:20.773 --> 07:23.670
I mean the host has the IP address of .20.

181
07:23.670 --> 07:26.100
And let's also imagine that we have our gateways

182
07:26.100 --> 07:28.710
fully configured, and we have connectivity out

183
07:28.710 --> 07:31.350
to the external networks, including the internet.
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

183
07:28.710 --> 07:31.350
to the external networks, including the internet.

184
07:31.350 --> 07:33.600
So I'll go ahead and put the internet right here,

185
07:33.600 --> 07:36.120
and I'll have this red line right here

186
07:36.120 --> 07:37.920
reflect our transport network,

187
07:37.920 --> 07:40.350
which could include many switches and routers,

188
07:40.350 --> 07:42.060
but effectively allowing communications

189
07:42.060 --> 07:44.640
between the tunnel endpoint here on ESXi-A

190
07:44.640 --> 07:47.130
and the tunnel endpoint on one of our edges.

191
07:47.130 --> 07:49.320
So let's imagine our customers sitting here,

192
07:49.320 --> 07:51.510
and they wanna go out to some internet resource,

193
07:51.510 --> 07:53.010
like a Google DNS server

194
07:53.010 --> 07:54.930
or a public internet website, et cetera,

195
07:54.930 --> 07:56.610
and let's also imagine we've already set up

196
07:56.610 --> 07:59.580
the north south routing here inside of NSX.

197
07:59.580 --> 08:01.487
When that client sends its traffic, ESXi-A,
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

197
07:59.580 --> 08:01.487
When that client sends its traffic, ESXi-A,

198
08:01.487 --> 08:03.510
'cause it came in on an overlay segment,

199
08:03.510 --> 08:06.030
is gonna go ahead and forward that through the tunnel

200
08:06.030 --> 08:08.400
to one of the two edges.

201
08:08.400 --> 08:10.470
So I'll go ahead and just write edge here

202
08:10.470 --> 08:11.940
as the end point for this tunnel.

203
08:11.940 --> 08:14.400
And then the edge will deencapsulate that,

204
08:14.400 --> 08:16.290
and it can continue forwarding that packet

205
08:16.290 --> 08:17.850
in the direction it needs to go to.

206
08:17.850 --> 08:19.440
And so for just for clarity, let's imagine,

207
08:19.440 --> 08:22.380
it's Edge-1 that's supporting the gateway services

208
08:22.380 --> 08:24.750
and maybe Edge-2 is acting as the standby.

209
08:24.750 --> 08:27.180
So then ESXi-A with its tunnel here,

210
08:27.180 --> 08:29.160
we'd have a tunnel that goes up to Edge-1.

211
08:29.160 --> 08:30.540
So the customer's traffic would again,
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000

211
08:29.160 --> 08:30.540
So the customer's traffic would again,

212
08:30.540 --> 08:32.790
encapsulated go over the GENEVE tunnel.

213
08:32.790 --> 08:34.320
And then Edge-1 when it receives that traffic,

214
08:34.320 --> 08:35.520
would deencapsulate it,

215
08:35.520 --> 08:38.520
logically process it as it needs to regarding routing,

216
08:38.520 --> 08:40.410
and then forward it out to the next top

217
08:40.410 --> 08:41.910
out on the physical network.

218
08:41.910 --> 08:44.250
And based on this plan, on this topology,

219
08:44.250 --> 08:47.190
let's implement Edge-1 and Edge-2.

220
08:47.190 --> 08:49.081
And just as a reminder, we're gonna place Edge-1

221
08:49.081 --> 08:52.977
as a VM using the compute resources of ESXi-A

222
08:52.977 --> 08:56.793
and we'll have Edge-2 running on ESXi host B.