1
00:00:00,000 --> 00:00:00,944

2
00:00:00,944 --> 00:00:03,776
[AUDIO LOGO]

3
00:00:03,776 --> 00:00:07,090

4
00:00:07,090 --> 00:00:12,640
And now we're finally up to
the lowest path cost, which,

5
00:00:12,640 --> 00:00:17,530
of course, is rule number 4
in our OSPF path selection

6
00:00:17,530 --> 00:00:18,640
criteria.

7
00:00:18,640 --> 00:00:21,940
Now, I'm sure we've
all seen lowest path

8
00:00:21,940 --> 00:00:24,380
cost wins numerous times.

9
00:00:24,380 --> 00:00:26,470
So in this video,
we're not going

10
00:00:26,470 --> 00:00:28,990
to really be focusing
so much on proving

11
00:00:28,990 --> 00:00:31,160
that the lowest cost wins.

12
00:00:31,160 --> 00:00:34,840
We're going to be more focusing
on what is done with the cost

13
00:00:34,840 --> 00:00:38,740
automatically on Cisco
routers, how can we change

14
00:00:38,740 --> 00:00:40,520
what's done automatically.

15
00:00:40,520 --> 00:00:44,380
And we'll take a quick look at
setting those costs manually

16
00:00:44,380 --> 00:00:46,570
just to make sure we're
clear on how to do it

17
00:00:46,570 --> 00:00:49,600
even though we have done
it on a few of our paths

18
00:00:49,600 --> 00:00:52,280
already in our lab environment.

19
00:00:52,280 --> 00:00:55,690
So that said, once again,
let's take a quick look

20
00:00:55,690 --> 00:00:58,070
at our topology
although, in this case,

21
00:00:58,070 --> 00:01:00,850
we're going to be doing most
of our work on router 8.

22
00:01:00,850 --> 00:01:03,100
And then, of course, we'll
jump into the command line

23
00:01:03,100 --> 00:01:08,970
and take a look at how cost
works on the Cisco routers.

24
00:01:08,970 --> 00:01:10,640
So we're going to
be starting off

25
00:01:10,640 --> 00:01:14,570
by examining the existing
cost on an interface.

26
00:01:14,570 --> 00:01:16,670
And as I said during
the introduction,

27
00:01:16,670 --> 00:01:19,310
we will be focusing on
router 8, particularly

28
00:01:19,310 --> 00:01:22,080
on this gigabit-1 interface.

29
00:01:22,080 --> 00:01:25,940
And what we're going
to see is the defaults

30
00:01:25,940 --> 00:01:28,700
for setting the
cost automatically.

31
00:01:28,700 --> 00:01:31,310
And then we're going
to change that.

32
00:01:31,310 --> 00:01:35,570
This is handled by a command
called auto-cost reference

33
00:01:35,570 --> 00:01:36,410
bandwidth.

34
00:01:36,410 --> 00:01:40,730
And the default value
for this is actually 100.

35
00:01:40,730 --> 00:01:42,920
Now, the interesting
thing about this

36
00:01:42,920 --> 00:01:46,160
is what that 100 actually means.

37
00:01:46,160 --> 00:01:49,250
100 actually means
that the bandwidth

38
00:01:49,250 --> 00:01:54,110
is going to be calculated
as 10 to the eighth divided

39
00:01:54,110 --> 00:01:58,620
by the interface bandwidth
in bits per second.

40
00:01:58,620 --> 00:02:01,430
Now, given these
defaults, 10 to the eighth

41
00:02:01,430 --> 00:02:03,300
is actually 100 million.

42
00:02:03,300 --> 00:02:05,480
So that would be
100 million bits

43
00:02:05,480 --> 00:02:07,820
per second, which
is what we generally

44
00:02:07,820 --> 00:02:14,150
call 100 megabits per second
or, of course, fast ethernet.

45
00:02:14,150 --> 00:02:17,660
So this corresponds
to the number

46
00:02:17,660 --> 00:02:20,390
after the auto-cost
reference bandwidth.

47
00:02:20,390 --> 00:02:25,370
Now, the big problem with this
is that we can't go to decimals

48
00:02:25,370 --> 00:02:28,490
and we can't go to fractions.

49
00:02:28,490 --> 00:02:33,170
The unfortunate side effect
of that is, fast ethernet

50
00:02:33,170 --> 00:02:38,640
has a cost of 1 because it's 100
million divided by 100 million.

51
00:02:38,640 --> 00:02:42,020
However, since we can't go
to fractions in decimals,

52
00:02:42,020 --> 00:02:45,970
that means the gigabit is
also going to be a cost of 1.

53
00:02:45,970 --> 00:02:48,200
10 gig, a cost of 1.

54
00:02:48,200 --> 00:02:51,600
100 gig, 400 gig, and so on.

55
00:02:51,600 --> 00:02:55,340
So this number is going to
have to be adjusted to account

56
00:02:55,340 --> 00:02:57,133
for higher speed interfaces.

57
00:02:57,133 --> 00:02:58,550
And that's one of
the things we're

58
00:02:58,550 --> 00:03:01,500
going to be looking
at here in this video.

59
00:03:01,500 --> 00:03:03,530
Now, of course, as
we go along and we

60
00:03:03,530 --> 00:03:05,600
change to different
values, we'll,

61
00:03:05,600 --> 00:03:09,000
of course, be observing
what's happening here.

62
00:03:09,000 --> 00:03:12,320
So let's jump over to the
command line on router 8.

63
00:03:12,320 --> 00:03:13,940
Let's look at some
of the defaults.

64
00:03:13,940 --> 00:03:15,770
And then we'll start
changing things

65
00:03:15,770 --> 00:03:18,400
and see what the results are.

66
00:03:18,400 --> 00:03:21,060
So if we start off by
looking at our interface--

67
00:03:21,060 --> 00:03:24,930
we'll just say show IP
OSPF interface gigabit 1--

68
00:03:24,930 --> 00:03:29,490
we can see that, as predicted,
this has a cost of 1.

69
00:03:29,490 --> 00:03:31,740
And we can also take
a look at that default

70
00:03:31,740 --> 00:03:34,980
value of the auto-cost
reference bandwidth.

71
00:03:34,980 --> 00:03:39,890
If we say show run all
and then we include--

72
00:03:39,890 --> 00:03:41,300
we'll just do auto-cost.

73
00:03:41,300 --> 00:03:44,990
We can see that that default
value of auto-cost reference

74
00:03:44,990 --> 00:03:46,520
bandwidth is 100.

75
00:03:46,520 --> 00:03:50,510
So now let's see what happens if
we start changing the bandwidth

76
00:03:50,510 --> 00:03:52,820
on this interface.

77
00:03:52,820 --> 00:03:56,680
Now, as we start changing the
bandwidth on this interface

78
00:03:56,680 --> 00:03:59,500
to see how it affects
the calculation

79
00:03:59,500 --> 00:04:02,200
of the auto-cost
reference bandwidth,

80
00:04:02,200 --> 00:04:05,380
keep in mind that
bandwidth is just

81
00:04:05,380 --> 00:04:08,710
a reference number used
by our routing protocols,

82
00:04:08,710 --> 00:04:10,870
in this case, OSPF.

83
00:04:10,870 --> 00:04:13,250
EIGRP looks at it as well.

84
00:04:13,250 --> 00:04:15,940
But it doesn't literally
change the throughput

85
00:04:15,940 --> 00:04:17,089
of the interface.

86
00:04:17,089 --> 00:04:19,870
So although this is still
a gigabit interface,

87
00:04:19,870 --> 00:04:23,870
we can have our routing protocol
treat it as if it's not.

88
00:04:23,870 --> 00:04:26,080
And that's what we're
going to do in order

89
00:04:26,080 --> 00:04:29,600
to look at how auto-cost
reference bandwidth functions.

90
00:04:29,600 --> 00:04:31,930
So let's take a look.

91
00:04:31,930 --> 00:04:37,600
We'll go into configuration
mode for interface gigabit 1.

92
00:04:37,600 --> 00:04:39,790
And if we type
"bandwidth," notice

93
00:04:39,790 --> 00:04:41,860
that if we're going
to hard-code a number,

94
00:04:41,860 --> 00:04:44,810
this is in kilobits per second.

95
00:04:44,810 --> 00:04:49,340
So remember, OSPF is looking
at this in bits per second.

96
00:04:49,340 --> 00:04:53,950
So these numbers are going
to be off by, well, 1,000.

97
00:04:53,950 --> 00:04:56,470
So let's go ahead and
change the bandwidth here

98
00:04:56,470 --> 00:05:01,150
to 100,000, which is going
to be the equivalent of 100

99
00:05:01,150 --> 00:05:02,780
megabits per second.

100
00:05:02,780 --> 00:05:07,240
And if we do our show IP OSPF
interface for gigabit 1 again,

101
00:05:07,240 --> 00:05:10,120
we see that the
cost hasn't changed

102
00:05:10,120 --> 00:05:14,410
because this is still 10 to the
eighth over 10 to the eighth.

103
00:05:14,410 --> 00:05:18,370
So here, we can see that
this 100-megabit-per-second

104
00:05:18,370 --> 00:05:22,000
interface has the
same cost as gigabit.

105
00:05:22,000 --> 00:05:24,550
But what if we were to change
the bandwidth one more time

106
00:05:24,550 --> 00:05:27,910
and just drop one 0,
which would effectively

107
00:05:27,910 --> 00:05:32,530
make this 10 megabits per second
or the equivalent of legacy

108
00:05:32,530 --> 00:05:33,400
ethernet?

109
00:05:33,400 --> 00:05:35,050
We'll look at our
interface again.

110
00:05:35,050 --> 00:05:37,940
Notice that now the cost is 10.

111
00:05:37,940 --> 00:05:41,720
So once we get down to
10 megabits per second,

112
00:05:41,720 --> 00:05:45,080
now the cost starts to
actually do something.

113
00:05:45,080 --> 00:05:49,180
So as we can see here,
the default behavior

114
00:05:49,180 --> 00:05:54,980
of auto-cost reference bandwidth
isn't ideal on most platforms.

115
00:05:54,980 --> 00:05:56,830
Now, one quick note on this.

116
00:05:56,830 --> 00:05:59,710
Cisco has begun to
change the default

117
00:05:59,710 --> 00:06:03,010
value of auto-cost
reference bandwidth

118
00:06:03,010 --> 00:06:05,170
on some of the newer platforms.

119
00:06:05,170 --> 00:06:07,690
So be sure to check
on whatever routers

120
00:06:07,690 --> 00:06:11,080
you are dealing with to find out
what the default value of this

121
00:06:11,080 --> 00:06:11,800
is.

122
00:06:11,800 --> 00:06:14,830
It may actually be
set appropriately high

123
00:06:14,830 --> 00:06:18,970
on more modern equipment
or higher-end equipment.

124
00:06:18,970 --> 00:06:23,620
As an example, NX-OS
on the Nexus platforms

125
00:06:23,620 --> 00:06:26,770
actually sets this to
a much higher number.

126
00:06:26,770 --> 00:06:28,690
Now, before we
change the auto-cost,

127
00:06:28,690 --> 00:06:32,110
let's go ahead and take our
artificial bandwidth number off

128
00:06:32,110 --> 00:06:33,700
of this interface.

129
00:06:33,700 --> 00:06:35,710
Then let's go over to OSPF.

130
00:06:35,710 --> 00:06:38,067
And we'll say auto-cost
reference bandwidth.

131
00:06:38,067 --> 00:06:39,650
And if you hit the
question mark here,

132
00:06:39,650 --> 00:06:42,400
we can see the value
range for this.

133
00:06:42,400 --> 00:06:45,730
And we do put this in,
in megabits per second.

134
00:06:45,730 --> 00:06:47,530
And as you can tell,
that fits right

135
00:06:47,530 --> 00:06:51,010
into the example we
did on our topology

136
00:06:51,010 --> 00:06:53,680
where we said that the
100 is the same as the 100

137
00:06:53,680 --> 00:06:55,900
megabits per second
for fast ethernet

138
00:06:55,900 --> 00:06:58,760
because this is in
megabits per second.

139
00:06:58,760 --> 00:07:00,490
Again, the default is 100.

140
00:07:00,490 --> 00:07:04,950
So what would happen if we made
this, let's say, a million?

141
00:07:04,950 --> 00:07:06,900
So it tells us that
the reference bandwidth

142
00:07:06,900 --> 00:07:08,400
has been changed.

143
00:07:08,400 --> 00:07:12,600
And it tells us to make this
consistent across all routers.

144
00:07:12,600 --> 00:07:14,610
The reason it does
this is, having

145
00:07:14,610 --> 00:07:18,660
a different view of the network
and different interpretation

146
00:07:18,660 --> 00:07:22,140
of costs can create
looping problems

147
00:07:22,140 --> 00:07:25,090
in routing protocols in general.

148
00:07:25,090 --> 00:07:26,880
So if you're going
to change metrics,

149
00:07:26,880 --> 00:07:29,335
it's always a good idea to
change it on everything.

150
00:07:29,335 --> 00:07:31,710
So now that we've changed it,
let's look at our interface

151
00:07:31,710 --> 00:07:32,310
again.

152
00:07:32,310 --> 00:07:35,670
And we see now that we
have a cost of 1,000.

153
00:07:35,670 --> 00:07:39,030
This is much more reasonable
because this would clearly

154
00:07:39,030 --> 00:07:42,510
be able to handle at
least 10-gig interfaces,

155
00:07:42,510 --> 00:07:46,510
100-gig interfaces,
400-gig interfaces.

156
00:07:46,510 --> 00:07:49,560
So this would probably
cover us for quite some time

157
00:07:49,560 --> 00:07:51,820
before we had to
adjust it again.

158
00:07:51,820 --> 00:07:54,900
So this isn't
necessarily a bad value.

159
00:07:54,900 --> 00:07:57,510
But one thing that we
do need to keep in mind

160
00:07:57,510 --> 00:08:01,110
is, as we turn up the
auto-cost reference bandwidth

161
00:08:01,110 --> 00:08:04,350
to handle these
faster interfaces,

162
00:08:04,350 --> 00:08:08,010
well, the problem that we had
with the high-speed interfaces

163
00:08:08,010 --> 00:08:11,340
before all having
a cost of 1, well,

164
00:08:11,340 --> 00:08:14,940
that exact same problem
is now going to exist

165
00:08:14,940 --> 00:08:17,580
on the lower-speed interfaces.

166
00:08:17,580 --> 00:08:21,960
Those are all now going to
go to maximum value, which,

167
00:08:21,960 --> 00:08:25,240
depending on your environment,
may or may not be a problem.

168
00:08:25,240 --> 00:08:28,800
Maybe you don't have any
slow-speed interfaces anymore

169
00:08:28,800 --> 00:08:31,500
in which case you don't
have to worry about this.

170
00:08:31,500 --> 00:08:34,870
But let's take a look at
what happens if we do.

171
00:08:34,870 --> 00:08:37,020
So let's go back into
our interface again.

172
00:08:37,020 --> 00:08:39,990
And let's change this
back to our 10 megs.

173
00:08:39,990 --> 00:08:42,820

174
00:08:42,820 --> 00:08:48,890
Notice the cost here has gone
to 65,535, the maximum value.

175
00:08:48,890 --> 00:08:52,390
And if I were to make
this ridiculously slow,

176
00:08:52,390 --> 00:08:55,630
let's drop it all the way
down to a good, old 64k link.

177
00:08:55,630 --> 00:08:58,590

178
00:08:58,590 --> 00:08:59,550
Same thing.

179
00:08:59,550 --> 00:09:03,210
See, the cost is still 65,535.

180
00:09:03,210 --> 00:09:07,890
So what's happened now is,
we've lost the resolution

181
00:09:07,890 --> 00:09:09,940
on the lower-speed interfaces.

182
00:09:09,940 --> 00:09:14,310
Now, the lower-speed interfaces
all look alike to OSPF.

183
00:09:14,310 --> 00:09:18,150
Now, this, of course, could
be very easily fixed by you

184
00:09:18,150 --> 00:09:19,470
as the administrator.

185
00:09:19,470 --> 00:09:21,750
If you did have this
problem, you could just

186
00:09:21,750 --> 00:09:25,170
say IP OSPF cost, let's say.

187
00:09:25,170 --> 00:09:27,780
And by the way, the maximum
for any individual interface

188
00:09:27,780 --> 00:09:35,290
is that 65,535, but we
could, say, set it to 50,000.

189
00:09:35,290 --> 00:09:37,540
So we can see now that
we've manually set it.

190
00:09:37,540 --> 00:09:42,340
So all we've done is override
what the auto-cost reference

191
00:09:42,340 --> 00:09:44,810
bandwidth was trying
to do on its own.

192
00:09:44,810 --> 00:09:47,800
Now, keep in mind that's
not necessarily a problem.

193
00:09:47,800 --> 00:09:49,720
If I turn this back off--

194
00:09:49,720 --> 00:09:52,520
we'll just go back to
auto-cost reference bandwidth.

195
00:09:52,520 --> 00:09:56,470
So once again, we're
just back to the 65,535.

196
00:09:56,470 --> 00:10:00,040
This does not in any way
prevent routes from being used.

197
00:10:00,040 --> 00:10:03,220
If I do a show IP
route OSPF, we're

198
00:10:03,220 --> 00:10:05,020
still using all these routes.

199
00:10:05,020 --> 00:10:08,290
Now, they're barely
over the 65,535

200
00:10:08,290 --> 00:10:09,910
because the rest
of our environment

201
00:10:09,910 --> 00:10:14,510
still has all 1s for all of
those gigabit interfaces.

202
00:10:14,510 --> 00:10:17,440
So as suggested,
we really should

203
00:10:17,440 --> 00:10:19,900
change the auto-cost
reference bandwidth

204
00:10:19,900 --> 00:10:22,070
on all of our devices.

205
00:10:22,070 --> 00:10:25,150
So let's hop over to router 1.

206
00:10:25,150 --> 00:10:28,180
We simply go into router OSPF 1.

207
00:10:28,180 --> 00:10:30,610
And we set the auto-cost
reference bandwidth

208
00:10:30,610 --> 00:10:31,690
to 1 million.

209
00:10:31,690 --> 00:10:33,010
Commit the changes.

210
00:10:33,010 --> 00:10:35,710
And we could verify
one of our interfaces.

211
00:10:35,710 --> 00:10:39,130
And we see here that
just like on IOS XE,

212
00:10:39,130 --> 00:10:41,320
the cost is now 1,000.

213
00:10:41,320 --> 00:10:44,290
You might have also noticed
that the command is pretty much

214
00:10:44,290 --> 00:10:44,962
identical.

215
00:10:44,962 --> 00:10:47,170
So we're going to configure
the rest of these routers

216
00:10:47,170 --> 00:10:50,170
to all have the same
auto-cost reference bandwidth,

217
00:10:50,170 --> 00:10:51,940
but we'll do it very quickly.

218
00:10:51,940 --> 00:10:58,048
There's R2, R3, R4, R5, 6, 7.

219
00:10:58,048 --> 00:10:59,200
8 is done.

220
00:10:59,200 --> 00:11:01,510
And 9 is not running OSPF.

221
00:11:01,510 --> 00:11:02,980
So that just leaves 10.

222
00:11:02,980 --> 00:11:06,820
Now all of our costs
should reference 1,000

223
00:11:06,820 --> 00:11:08,320
on all the interfaces.

224
00:11:08,320 --> 00:11:10,900
We can go back and
verify this on router 8

225
00:11:10,900 --> 00:11:14,550
since that's the router
we've been working on.

226
00:11:14,550 --> 00:11:17,340
So if we just generically look
at the routing table again,

227
00:11:17,340 --> 00:11:23,400
we see that these are all
values like 67,535 and so on.

228
00:11:23,400 --> 00:11:25,280
Now, this, of course,
is because we still

229
00:11:25,280 --> 00:11:28,010
have that ridiculously
low bandwidth

230
00:11:28,010 --> 00:11:29,720
on our gigabit-1 interface.

231
00:11:29,720 --> 00:11:30,920
Let's go ahead and fix that.

232
00:11:30,920 --> 00:11:34,670

233
00:11:34,670 --> 00:11:38,030
We'll just say no bandwidth, and
it will go back to its default.

234
00:11:38,030 --> 00:11:40,010
We'll look at the routing
table one more time.

235
00:11:40,010 --> 00:11:43,760
And we can see that these are
more reasonable values given

236
00:11:43,760 --> 00:11:45,900
that all of our interfaces
in our environment

237
00:11:45,900 --> 00:11:47,810
now count for 1,000.

238
00:11:47,810 --> 00:11:50,120
You may notice, by the
way, as a side note

239
00:11:50,120 --> 00:11:53,570
that a lot of these
values end in a 1,

240
00:11:53,570 --> 00:11:56,570
such as let's pick on
router 1's loopback here

241
00:11:56,570 --> 00:11:59,180
with a cost of 2,001.

242
00:11:59,180 --> 00:12:03,230
That's because that by default,
loopback interfaces always

243
00:12:03,230 --> 00:12:04,700
have a cost of 1.

244
00:12:04,700 --> 00:12:07,640
And that is a route to
router 1's loopback.

245
00:12:07,640 --> 00:12:11,060
So it's 2,000 to get
to router 1 and then

246
00:12:11,060 --> 00:12:14,350
1 to get to its loopback.

247
00:12:14,350 --> 00:12:19,810
So in this video, we really
focused in on the cost value.

248
00:12:19,810 --> 00:12:22,360
We saw how auto-cost
reference bandwidth

249
00:12:22,360 --> 00:12:25,930
works, how it calculates
the default bandwidth

250
00:12:25,930 --> 00:12:29,830
for each interface based
on an inverse proportion

251
00:12:29,830 --> 00:12:31,400
of its bandwidth.

252
00:12:31,400 --> 00:12:35,230
We also saw that we can manually
set the cost on an interface,

253
00:12:35,230 --> 00:12:37,840
of course, to
manipulate routing.

254
00:12:37,840 --> 00:12:41,320
You could really
change the bandwidth

255
00:12:41,320 --> 00:12:44,650
to manipulate routing as well,
although that's generally

256
00:12:44,650 --> 00:12:47,860
not recommended because
other things actually

257
00:12:47,860 --> 00:12:52,240
do look at that bandwidth
value, such as QoS.

258
00:12:52,240 --> 00:12:55,990
So it's really recommended
that you change the cost

259
00:12:55,990 --> 00:12:58,810
if you want to manipulate OSPF.

260
00:12:58,810 --> 00:13:01,840
So don't forget that on
most of our platforms,

261
00:13:01,840 --> 00:13:04,390
the value of auto-cost
reference bandwidth

262
00:13:04,390 --> 00:13:09,280
defaults to 100, which is the
equivalent of fast ethernet.

263
00:13:09,280 --> 00:13:13,450
So this does need to be changed
on most of today's devices.

264
00:13:13,450 --> 00:13:16,690
But keep in mind Cisco is
beginning to change that

265
00:13:16,690 --> 00:13:18,500
on some of their platforms.

266
00:13:18,500 --> 00:13:22,910
So always watch what your
platform defaults to.

267
00:13:22,910 --> 00:13:25,000
I hope this has been
informative for you.

268
00:13:25,000 --> 00:13:28,110
And I'd like to thank
you for viewing.

269
00:13:28,110 --> 00:13:31,000