1
00:00:00,000 --> 00:00:00,960
In this lesson,

2
00:00:00,960 --> 00:00:03,840
we're going to discuss some high availability approaches.

3
00:00:03,840 --> 00:00:05,880
Now, it's essential to achieve high availability

4
00:00:05,880 --> 00:00:07,200
in your network environments

5
00:00:07,200 --> 00:00:09,179
if you want to be able to maintain continuous operations

6
00:00:09,179 --> 00:00:12,150
with minimal downtime for your end users.

7
00:00:12,150 --> 00:00:14,490
Network redundancy is focused on ensuring our networks

8
00:00:14,490 --> 00:00:16,347
remain up and running at all times

9
00:00:16,347 --> 00:00:18,660
so that we can increase our availability.

10
00:00:18,660 --> 00:00:20,790
This includes the server connections to the network,

11
00:00:20,790 --> 00:00:22,800
the connections between our switches and our routers

12
00:00:22,800 --> 00:00:24,720
and our connections out to the internet.

13
00:00:24,720 --> 00:00:25,890
To accomplish all of this,

14
00:00:25,890 --> 00:00:28,680
our servers have two or more network interface parts,

15
00:00:28,680 --> 00:00:30,180
and they can be operated as a pair

16
00:00:30,180 --> 00:00:32,580
or in a load balancing configuration.

17
00:00:32,580 --> 00:00:34,890
For example, I can have a network attached server

18
00:00:34,890 --> 00:00:36,840
with five network cards installed.

19
00:00:36,840 --> 00:00:37,870
On this particular server,

20
00:00:37,870 --> 00:00:40,380
I have four network cables plugged into it,

21
00:00:40,380 --> 00:00:42,210
and it's already providing services.

22
00:00:42,210 --> 00:00:44,631
Now under this configuration, if one of those cards fails,

23
00:00:44,631 --> 00:00:47,610
the other ones can take over the load and the demand.

24
00:00:47,610 --> 00:00:49,890
Now, these network cards can also be used in pairs

25
00:00:49,890 --> 00:00:50,970
for more redundancy,

26
00:00:50,970 --> 00:00:53,580
or we can split them up and put them into two groups

27
00:00:53,580 --> 00:00:54,750
so you have additional throughput

28
00:00:54,750 --> 00:00:57,240
by operating more of these cards at one time

29
00:00:57,240 --> 00:00:59,010
in a simultaneous manner.

30
00:00:59,010 --> 00:01:00,783
So let's say I create two groups,

31
00:01:00,783 --> 00:01:02,550
one with four network cards in it,

32
00:01:02,550 --> 00:01:05,610
and one with a single network card over there by itself.

33
00:01:05,610 --> 00:01:06,690
Now this will give me a total

34
00:01:06,690 --> 00:01:09,270
of five different network connections that we can have.

35
00:01:09,270 --> 00:01:11,970
Four of those can operate in a shared load capacity,

36
00:01:11,970 --> 00:01:13,020
and this means that all four

37
00:01:13,020 --> 00:01:15,600
will work together as one virtual card.

38
00:01:15,600 --> 00:01:18,810
So if each one is 100 megabits per second each,

39
00:01:18,810 --> 00:01:20,340
this will give me a combined amount

40
00:01:20,340 --> 00:01:22,410
of 400 megabits per second,

41
00:01:22,410 --> 00:01:24,210
and that fifth one sitting over there by itself,

42
00:01:24,210 --> 00:01:26,970
it will still be 100 megabits per second by itself,

43
00:01:26,970 --> 00:01:28,890
and we can use that as a backup connection

44
00:01:28,890 --> 00:01:31,530
for increased levels of redundancy if we need to.

45
00:01:31,530 --> 00:01:33,814
Now, if the group of four connections happens to go offline,

46
00:01:33,814 --> 00:01:36,018
that fifth one sitting by itself can take over

47
00:01:36,018 --> 00:01:37,950
and provide some level of service,

48
00:01:37,950 --> 00:01:40,410
even though it's going to be a slower connection.

49
00:01:40,410 --> 00:01:42,546
Now, in order to ensure our network devices have redundancy,

50
00:01:42,546 --> 00:01:44,550
we're also going to use things like switches

51
00:01:44,550 --> 00:01:45,780
and routers that are configured

52
00:01:45,780 --> 00:01:47,490
with multiple and redundant network cables

53
00:01:47,490 --> 00:01:48,810
going between them.

54
00:01:48,810 --> 00:01:50,370
For example, if I have a switch

55
00:01:50,370 --> 00:01:52,740
with two trunk ports using fiber optic cables,

56
00:01:52,740 --> 00:01:54,180
this can allow me to have that switch

57
00:01:54,180 --> 00:01:56,640
to be able to get service from two separate routers.

58
00:01:56,640 --> 00:01:58,591
And this way, if one of the routers goes down,

59
00:01:58,591 --> 00:02:01,500
the other router can take over as the secondary one,

60
00:02:01,500 --> 00:02:03,074
and it'll carry all the data from that switch

61
00:02:03,074 --> 00:02:04,650
out to the network.

62
00:02:04,650 --> 00:02:06,150
This will then give me redundancy

63
00:02:06,150 --> 00:02:08,820
by allowing us to have different pathways out of that switch

64
00:02:08,820 --> 00:02:11,160
by using these two different network connections.

65
00:02:11,160 --> 00:02:11,993
So as you can see,

66
00:02:11,993 --> 00:02:13,589
when you're planning for network redundancy,

67
00:02:13,589 --> 00:02:15,570
you really need to start thinking about this

68
00:02:15,570 --> 00:02:17,280
in three different parts.

69
00:02:17,280 --> 00:02:18,649
You need to think about it as the device,

70
00:02:18,649 --> 00:02:21,734
those network interface cards and the cables themself.

71
00:02:21,734 --> 00:02:23,970
Also, you need to think about it from the router

72
00:02:23,970 --> 00:02:26,370
and switch perspective to make sure you have redundant paths

73
00:02:26,370 --> 00:02:28,800
inside of your network and out to the internet as well

74
00:02:28,800 --> 00:02:31,260
when you're making external connections.

75
00:02:31,260 --> 00:02:33,030
All right, now that we've covered some of the basics,

76
00:02:33,030 --> 00:02:35,400
let's explore some different high availability approaches

77
00:02:35,400 --> 00:02:37,170
or strategies that we can utilize

78
00:02:37,170 --> 00:02:38,790
to achieve higher levels of availability

79
00:02:38,790 --> 00:02:40,680
and redundancy in our networks.

80
00:02:40,680 --> 00:02:43,050
This includes things like active-active approaches,

81
00:02:43,050 --> 00:02:44,242
active-passive approaches,

82
00:02:44,242 --> 00:02:46,740
the use of hardware devices like load balancers,

83
00:02:46,740 --> 00:02:49,110
and the use of content delivery networks.

84
00:02:49,110 --> 00:02:50,400
First, let's explore

85
00:02:50,400 --> 00:02:53,220
the active-active approach to high availability.

86
00:02:53,220 --> 00:02:55,140
Now, inside of an active-active setup,

87
00:02:55,140 --> 00:02:57,102
multiple systems will be run simultaneously

88
00:02:57,102 --> 00:02:59,730
and they'll share the load and provide seamless service

89
00:02:59,730 --> 00:03:02,010
even if one or more of those systems fail.

90
00:03:02,010 --> 00:03:04,560
This approach will maximize resource utilization

91
00:03:04,560 --> 00:03:06,720
and ensure that the service can handle the load

92
00:03:06,720 --> 00:03:09,000
even in a case of a system failure.

93
00:03:09,000 --> 00:03:11,130
For example, if I have two network cards

94
00:03:11,130 --> 00:03:13,110
installed in my network-attached file server,

95
00:03:13,110 --> 00:03:15,306
and each card operates at one gigabit per second,

96
00:03:15,306 --> 00:03:17,670
I can put them in a load balance configuration

97
00:03:17,670 --> 00:03:19,830
that supports the active-active configuration.

98
00:03:19,830 --> 00:03:21,604
And that way, if both cards are operating,

99
00:03:21,604 --> 00:03:23,130
they'll be working together

100
00:03:23,130 --> 00:03:26,520
to provide a combined throughput of two gigabits per second.

101
00:03:26,520 --> 00:03:27,960
But if one of those network connections

102
00:03:27,960 --> 00:03:29,280
goes down for any reason,

103
00:03:29,280 --> 00:03:31,290
the other one is already there and working.

104
00:03:31,290 --> 00:03:33,629
So it's going to continue to work because it's already active.

105
00:03:33,629 --> 00:03:35,970
The only difference that we're going to see as an end user

106
00:03:35,970 --> 00:03:38,400
is that our speed or our throughput has gone down

107
00:03:38,400 --> 00:03:39,840
because it's been cut in half

108
00:03:39,840 --> 00:03:41,880
because we lost one of those two connectionsm

109
00:03:41,880 --> 00:03:43,500
so we went from one gigabit per second

110
00:03:43,500 --> 00:03:44,910
plus one gigabit per second,

111
00:03:44,910 --> 00:03:46,880
or two gigabits per second of combined throughput

112
00:03:46,880 --> 00:03:49,980
down to only a single one gigabit per second connection.

113
00:03:49,980 --> 00:03:50,813
And that means we only have

114
00:03:50,813 --> 00:03:51,960
one gigabit per second of throughput

115
00:03:51,960 --> 00:03:53,663
that we can use for our systems.

116
00:03:53,663 --> 00:03:55,530
Now the second thing we need to discuss

117
00:03:55,530 --> 00:03:58,530
is an active-passive approach to high availability.

118
00:03:58,530 --> 00:04:00,510
When we're using an active-passive approach,

119
00:04:00,510 --> 00:04:01,830
our systems are going to be configured

120
00:04:01,830 --> 00:04:04,050
so that a standby system will remain idle

121
00:04:04,050 --> 00:04:06,090
until the primary system fails.

122
00:04:06,090 --> 00:04:08,250
When the primary or active system fails,

123
00:04:08,250 --> 00:04:10,110
then our standby system will go in

124
00:04:10,110 --> 00:04:12,090
and take over to ensure that the continuity

125
00:04:12,090 --> 00:04:15,060
of our services will remain available and online.

126
00:04:15,060 --> 00:04:16,740
Now, one thing to note about this approach

127
00:04:16,740 --> 00:04:18,360
is that it may not utilize resources

128
00:04:18,360 --> 00:04:21,029
as efficiently as an active-active setup,

129
00:04:21,029 --> 00:04:22,230
but it does still provide

130
00:04:22,230 --> 00:04:25,080
a reliable fallback mechanism for us to use.

131
00:04:25,080 --> 00:04:26,460
Let me give you a real-world example

132
00:04:26,460 --> 00:04:29,190
of an active-active and active-passive setup.

133
00:04:29,190 --> 00:04:30,540
Now at my old office building,

134
00:04:30,540 --> 00:04:31,500
our power was configured

135
00:04:31,500 --> 00:04:33,450
to use an active standby configuration

136
00:04:33,450 --> 00:04:34,830
when we first moved in.

137
00:04:34,830 --> 00:04:36,420
Because our offices were located in a region

138
00:04:36,420 --> 00:04:38,850
where hurricanes and power outages occurred often,

139
00:04:38,850 --> 00:04:41,970
the previous owner had installed a large diesel generator.

140
00:04:41,970 --> 00:04:43,890
Anytime the power to the building was lost,

141
00:04:43,890 --> 00:04:45,300
the generator would detect that,

142
00:04:45,300 --> 00:04:46,350
it would start itself up

143
00:04:46,350 --> 00:04:47,670
in response to the power loss,

144
00:04:47,670 --> 00:04:49,620
and after about 30 to 60 seconds,

145
00:04:49,620 --> 00:04:52,050
it would come online and take over providing power

146
00:04:52,050 --> 00:04:53,400
until the primary power source

147
00:04:53,400 --> 00:04:55,192
was available for us to use again.

148
00:04:55,192 --> 00:04:56,370
Now, unfortunately,

149
00:04:56,370 --> 00:04:58,230
this left us in the dark for about a minute

150
00:04:58,230 --> 00:04:59,820
while that generator was turning on

151
00:04:59,820 --> 00:05:01,470
and getting ready to assume the power loads

152
00:05:01,470 --> 00:05:02,460
to run the building,

153
00:05:02,460 --> 00:05:03,630
and we really didn't like this setup

154
00:05:03,630 --> 00:05:06,240
from a high availability or redundancy perspective,

155
00:05:06,240 --> 00:05:07,560
and so we decided to change it

156
00:05:07,560 --> 00:05:09,810
from this active-passive mode into something

157
00:05:09,810 --> 00:05:12,750
that uses an active-active configuration instead.

158
00:05:12,750 --> 00:05:13,583
To achieve this,

159
00:05:13,583 --> 00:05:15,300
we first installed 60 solar panels

160
00:05:15,300 --> 00:05:16,408
to generate enough electricity

161
00:05:16,408 --> 00:05:18,630
to be our primary power source.

162
00:05:18,630 --> 00:05:20,685
Then we installed an electrical transfer switch

163
00:05:20,685 --> 00:05:22,080
that would automatically go

164
00:05:22,080 --> 00:05:23,970
from the solar panels to the electrical grid

165
00:05:23,970 --> 00:05:25,530
anytime we're not getting enough power generated

166
00:05:25,530 --> 00:05:26,820
by those solar panels

167
00:05:26,820 --> 00:05:28,964
because maybe it's a cloudy day or something like that.

168
00:05:28,964 --> 00:05:32,190
This was effectively an active standby configuration,

169
00:05:32,190 --> 00:05:34,020
similar to what we discussed with a generator,

170
00:05:34,020 --> 00:05:35,490
because we would use all solar

171
00:05:35,490 --> 00:05:38,820
or we would use all electric power company provided grid,

172
00:05:38,820 --> 00:05:41,520
we wouldn't use both of them at the same time.

173
00:05:41,520 --> 00:05:44,010
Now, to achieve an active-active configuration,

174
00:05:44,010 --> 00:05:45,300
we actually went ahead and installed

175
00:05:45,300 --> 00:05:47,460
an always on battery backup system.

176
00:05:47,460 --> 00:05:48,293
By using this system,

177
00:05:48,293 --> 00:05:50,610
our solar panels would first fill the batteries,

178
00:05:50,610 --> 00:05:52,050
and then the batteries would discharge

179
00:05:52,050 --> 00:05:53,940
the power into the office building.

180
00:05:53,940 --> 00:05:55,448
If the solar went down for any reason,

181
00:05:55,448 --> 00:05:57,030
we actually wouldn't even notice it

182
00:05:57,030 --> 00:05:59,880
because the battery backup will continue to provide power

183
00:05:59,880 --> 00:06:01,620
for between four and seven days

184
00:06:01,620 --> 00:06:03,657
based on the system size that we had installed

185
00:06:03,657 --> 00:06:06,090
and our level of power utilization.

186
00:06:06,090 --> 00:06:07,290
So hopefully you can see the difference here

187
00:06:07,290 --> 00:06:09,019
between the active passive configuration

188
00:06:09,019 --> 00:06:11,550
where we are using one of two power sources

189
00:06:11,550 --> 00:06:13,590
with the ability to switch to the second one

190
00:06:13,590 --> 00:06:15,450
and an active-active configuration

191
00:06:15,450 --> 00:06:17,310
where we are constantly feeding the batteries,

192
00:06:17,310 --> 00:06:18,840
and then those batteries are feeding the power

193
00:06:18,840 --> 00:06:19,740
to our building,

194
00:06:19,740 --> 00:06:23,550
so we always can use both power sources at the same time.

195
00:06:23,550 --> 00:06:26,099
Now, third, let's take a look at load balancers.

196
00:06:26,099 --> 00:06:28,230
Load balancers are going to function inside

197
00:06:28,230 --> 00:06:30,240
of our high availability approaches as well,

198
00:06:30,240 --> 00:06:32,160
and they're going to be used to distribute network traffic

199
00:06:32,160 --> 00:06:33,870
across multiple servers to ensure

200
00:06:33,870 --> 00:06:35,959
that no single server will bear too much load,

201
00:06:35,959 --> 00:06:38,040
and this way we can maintain higher levels

202
00:06:38,040 --> 00:06:39,593
of performance and uptime.

203
00:06:39,593 --> 00:06:42,450
These load balancers really do play a crucial role

204
00:06:42,450 --> 00:06:43,950
in an active-active setup

205
00:06:43,950 --> 00:06:46,050
by efficiently distributing incoming requests

206
00:06:46,050 --> 00:06:48,060
to the best available resources.

207
00:06:48,060 --> 00:06:50,100
Load balancers also will continuously

208
00:06:50,100 --> 00:06:51,480
monitor the health of the servers,

209
00:06:51,480 --> 00:06:53,040
and they can reroute your traffic away

210
00:06:53,040 --> 00:06:56,310
from failed nodes to maintain uninterrupted services.

211
00:06:56,310 --> 00:06:58,529
Fourth, we have content delivery networks.

212
00:06:58,529 --> 00:07:00,711
Content delivery network or CDN

213
00:07:00,711 --> 00:07:02,220
is another vital component

214
00:07:02,220 --> 00:07:04,259
in achieving high levels of availability.

215
00:07:04,259 --> 00:07:06,900
A content delivery network is a network of servers

216
00:07:06,900 --> 00:07:08,204
that are distributed geographically

217
00:07:08,204 --> 00:07:12,000
to deliver content to users more efficiently and reliably.

218
00:07:12,000 --> 00:07:15,232
CDNs can also store cash content closer to the end users,

219
00:07:15,232 --> 00:07:16,740
so we can reduce the latency

220
00:07:16,740 --> 00:07:19,080
and lower the load times for those end users

221
00:07:19,080 --> 00:07:21,270
when they're trying to access those services.

222
00:07:21,270 --> 00:07:24,210
Now, in the case of a server failure or excessive traffic,

223
00:07:24,210 --> 00:07:25,246
our content delivery networks

224
00:07:25,246 --> 00:07:26,970
can simply reroute the request

225
00:07:26,970 --> 00:07:28,710
to the next closest server to ensure

226
00:07:28,710 --> 00:07:30,840
that our end users experience remains unaffected

227
00:07:30,840 --> 00:07:32,430
by the loss of a single server

228
00:07:32,430 --> 00:07:34,800
or higher than expected network traffic loads

229
00:07:34,800 --> 00:07:36,270
being experienced.

230
00:07:36,270 --> 00:07:38,393
Now, when it comes to implementing high availability,

231
00:07:38,393 --> 00:07:40,980
it's important that you design the network with redundancy

232
00:07:40,980 --> 00:07:41,813
at its core.

233
00:07:41,813 --> 00:07:43,380
Redundancy can be achieved

234
00:07:43,380 --> 00:07:45,300
by deploying multiple network interface cards

235
00:07:45,300 --> 00:07:46,560
inside of your servers,

236
00:07:46,560 --> 00:07:48,870
and this way, you can support both load balancing

237
00:07:48,870 --> 00:07:51,690
and redundancy by ensuring that multiple pathways exist

238
00:07:51,690 --> 00:07:53,190
between your switches and routers,

239
00:07:53,190 --> 00:07:55,140
as well as by having redundant internet connections

240
00:07:55,140 --> 00:07:56,460
for your use too.

241
00:07:56,460 --> 00:07:59,160
By combining these infrastructure setups with active-active

242
00:07:59,160 --> 00:08:01,920
or active-passive approaches and integrating load balancers

243
00:08:01,920 --> 00:08:03,900
and CDNs, your enterprise network

244
00:08:03,900 --> 00:08:06,570
can achieve high availability, minimize its downtime,

245
00:08:06,570 --> 00:08:08,760
and provide a seamless user experience.

246
00:08:08,760 --> 00:08:10,661
So remember, high availability approaches

247
00:08:10,661 --> 00:08:13,093
include things like active-active configurations

248
00:08:13,093 --> 00:08:15,330
and active-passive configuration,

249
00:08:15,330 --> 00:08:16,680
leveraging load balancers,

250
00:08:16,680 --> 00:08:18,990
or implementing a content delivery network

251
00:08:18,990 --> 00:08:20,880
so that you can maintain continuous operations

252
00:08:20,880 --> 00:08:22,410
and higher levels of performance

253
00:08:22,410 --> 00:08:24,570
in enterprise networks and services.

254
00:08:24,570 --> 00:08:26,310
These strategies can also be combined

255
00:08:26,310 --> 00:08:27,780
with a robust network design

256
00:08:27,780 --> 00:08:30,154
that can focus on integrating redundancy into your systems

257
00:08:30,154 --> 00:08:31,680
so that you can ensure your business

258
00:08:31,680 --> 00:08:33,780
can provide reliable and efficient services

259
00:08:33,780 --> 00:08:34,953
to its end users.

