1
00:00:00,000 --> 00:00:00,954

2
00:00:00,954 --> 00:00:04,293
[CBT NUGGETS THEME MUSIC]

3
00:00:04,293 --> 00:00:07,160

4
00:00:07,160 --> 00:00:09,950
So the next thing we want
to talk about with OSPF

5
00:00:09,950 --> 00:00:11,870
is optimizing it.

6
00:00:11,870 --> 00:00:14,030
And probably the
best way to optimize

7
00:00:14,030 --> 00:00:17,060
it is to start with looking
at some of its timers

8
00:00:17,060 --> 00:00:21,110
that we can adjust to make
it respond faster to errors

9
00:00:21,110 --> 00:00:22,950
and to converge faster.

10
00:00:22,950 --> 00:00:25,730
So let's start by jumping in and
looking at the different timers

11
00:00:25,730 --> 00:00:26,750
we can adjust.

12
00:00:26,750 --> 00:00:28,670
And then we'll jump
into the command line

13
00:00:28,670 --> 00:00:32,890
and see some of those
timers in action.

14
00:00:32,890 --> 00:00:35,550
So as we said, one
of the easiest ways--

15
00:00:35,550 --> 00:00:38,430
and you could argue probably
one of the best ways as well--

16
00:00:38,430 --> 00:00:42,750
to increase OSPF
performance is to adjust

17
00:00:42,750 --> 00:00:45,300
some of those default timers.

18
00:00:45,300 --> 00:00:47,670
Some of the default timers
are fine, and some of them

19
00:00:47,670 --> 00:00:48,850
need to be adjusted.

20
00:00:48,850 --> 00:00:50,550
So let's go through
and take a look

21
00:00:50,550 --> 00:00:53,310
at what we can
and should change.

22
00:00:53,310 --> 00:00:58,500
Keep in mind as we go through
this that the default timers

23
00:00:58,500 --> 00:01:00,360
for many protocols--

24
00:01:00,360 --> 00:01:02,550
I'm not just picking
on OSPF here--

25
00:01:02,550 --> 00:01:06,270
but of course,
including OSPF, these

26
00:01:06,270 --> 00:01:09,450
have been around for
a long time, decades.

27
00:01:09,450 --> 00:01:13,170
And at that time, we
had different networking

28
00:01:13,170 --> 00:01:17,640
technologies in place, and
possibly even more importantly,

29
00:01:17,640 --> 00:01:19,260
different equipment.

30
00:01:19,260 --> 00:01:22,200
I remember when I first
started doing Cisco routing.

31
00:01:22,200 --> 00:01:24,840
A lot of the routers
back then couldn't even

32
00:01:24,840 --> 00:01:27,510
run full-blown OSPF.

33
00:01:27,510 --> 00:01:31,020
They could run OSPF
if it was a stub area,

34
00:01:31,020 --> 00:01:34,320
but they couldn't
run full-blown OSPF.

35
00:01:34,320 --> 00:01:36,360
The equipment just
couldn't handle it.

36
00:01:36,360 --> 00:01:39,270
Now, of course, it's no problem.

37
00:01:39,270 --> 00:01:42,330
All of the devices can
run full-blown OSPF.

38
00:01:42,330 --> 00:01:44,760
And since the equipment
is so much better,

39
00:01:44,760 --> 00:01:49,590
that's going to change how
quickly they can process OSPF.

40
00:01:49,590 --> 00:01:53,500
A lot of these timers were
there to throttle these routers

41
00:01:53,500 --> 00:01:58,210
so that they could handle
OSPF on lesser equipment.

42
00:01:58,210 --> 00:02:01,140
So the first of these timers
that we should really look at

43
00:02:01,140 --> 00:02:05,430
are the timers that determine
how quickly we detect

44
00:02:05,430 --> 00:02:07,800
the failure of a neighbor.

45
00:02:07,800 --> 00:02:09,720
Now, the reason
we have to do this

46
00:02:09,720 --> 00:02:13,980
is to find a route around
that neighbor if possible,

47
00:02:13,980 --> 00:02:17,010
assuming that there
is redundant links.

48
00:02:17,010 --> 00:02:19,270
The two timers
involved in this--

49
00:02:19,270 --> 00:02:21,420
so the two things we have
to be concerned about--

50
00:02:21,420 --> 00:02:24,570
are the Hello and Dead timers.

51
00:02:24,570 --> 00:02:28,200
It is recommended to
keep these at a 4 to 1

52
00:02:28,200 --> 00:02:33,030
ratio, where the Dead timer
is four times the Hello timer.

53
00:02:33,030 --> 00:02:36,150
Now, note that the
default for these values

54
00:02:36,150 --> 00:02:40,110
is based on the
OSPF network type.

55
00:02:40,110 --> 00:02:43,530
Most modern networks--
this would include things

56
00:02:43,530 --> 00:02:46,500
like ethernet,
point-to-point lengths.

57
00:02:46,500 --> 00:02:48,750
So most of our
modern network types

58
00:02:48,750 --> 00:02:51,750
are going to default
to 10 and 40.

59
00:02:51,750 --> 00:02:56,910
Again, that's a 10-second Hello
with a 40-second Dead timer.

60
00:02:56,910 --> 00:02:59,700
Some traditional
WAN network types--

61
00:02:59,700 --> 00:03:02,730
and one that we might still have
to be concerned with today--

62
00:03:02,730 --> 00:03:05,100
would be network type
point-to-multipoint.

63
00:03:05,100 --> 00:03:07,620
The reason we still might
be concerned with this today

64
00:03:07,620 --> 00:03:11,790
is it does still fit on
technologies such as DMVPN.

65
00:03:11,790 --> 00:03:16,990
Honestly, OSPF isn't really
recommended on DMVPN anyway,

66
00:03:16,990 --> 00:03:20,070
but if you're going to run
it, point-to-multipoint

67
00:03:20,070 --> 00:03:22,150
might be a good
solution for you.

68
00:03:22,150 --> 00:03:26,160
However, we probably don't
want those default timers

69
00:03:26,160 --> 00:03:28,170
of 30 and 120.

70
00:03:28,170 --> 00:03:30,450
These timers were
traditionally used

71
00:03:30,450 --> 00:03:34,110
because these were supposed
to be WAN network types,

72
00:03:34,110 --> 00:03:36,450
and the WAN was
traditionally slower,

73
00:03:36,450 --> 00:03:40,320
so it was better to have slower
timers on the slower links.

74
00:03:40,320 --> 00:03:42,570
Now in case, it's
not clear by the fact

75
00:03:42,570 --> 00:03:45,150
that we're saying that the
timers are set by network type

76
00:03:45,150 --> 00:03:48,180
and such, that does
mean that the timers are

77
00:03:48,180 --> 00:03:51,190
going to be set per interface.

78
00:03:51,190 --> 00:03:53,400
So I can absolutely
have one interface

79
00:03:53,400 --> 00:03:56,010
set one way, one interface
set a different way

80
00:03:56,010 --> 00:03:57,180
on the same router.

81
00:03:57,180 --> 00:03:58,710
It's done all the time.

82
00:03:58,710 --> 00:04:01,770
These default values
that we mentioned here,

83
00:04:01,770 --> 00:04:04,680
these really should
rarely be used.

84
00:04:04,680 --> 00:04:08,460
When I say rarely, I
sort of mean never.

85
00:04:08,460 --> 00:04:10,800
You really shouldn't
use the default timers.

86
00:04:10,800 --> 00:04:14,070
I just don't really like to use
words like never because there

87
00:04:14,070 --> 00:04:18,180
may be a situation somewhere
where the default timers are

88
00:04:18,180 --> 00:04:20,279
still something
somebody wants to use.

89
00:04:20,279 --> 00:04:22,330
There's nothing wrong
with using them.

90
00:04:22,330 --> 00:04:24,720
It's just, again,
on today's networks,

91
00:04:24,720 --> 00:04:28,020
having 40 seconds of
downtime before we

92
00:04:28,020 --> 00:04:32,560
detect the failure of a neighbor
is almost never acceptable.

93
00:04:32,560 --> 00:04:34,230
So we're probably
hardly ever going

94
00:04:34,230 --> 00:04:36,510
to use these default timers.

95
00:04:36,510 --> 00:04:38,940
So now let's take a look at
some of the other timers we

96
00:04:38,940 --> 00:04:40,800
can adjust.

97
00:04:40,800 --> 00:04:44,420
So the other timers we're going
to look at have to do with how

98
00:04:44,420 --> 00:04:46,970
the Shortest Path
First algorithm is run,

99
00:04:46,970 --> 00:04:51,230
as well as how quickly or
how often we send link-state

100
00:04:51,230 --> 00:04:53,540
advertisements or LSAs--

101
00:04:53,540 --> 00:04:55,050
to our neighbors.

102
00:04:55,050 --> 00:05:00,480
Now, these timers can be
optimized either manually,

103
00:05:00,480 --> 00:05:02,780
which means you simply
go in and manually change

104
00:05:02,780 --> 00:05:05,660
the timers using the
commands in the IOS,

105
00:05:05,660 --> 00:05:08,810
or on most of today's
equipment, these

106
00:05:08,810 --> 00:05:12,770
will be set automatically
using a feature called

107
00:05:12,770 --> 00:05:16,190
routing-default-optimize.

108
00:05:16,190 --> 00:05:21,590
This is actually on by default,
starting in IOS XE version

109
00:05:21,590 --> 00:05:24,800
16.5.1b.

110
00:05:24,800 --> 00:05:28,340
It's also on by
default in IOS XR.

111
00:05:28,340 --> 00:05:31,040
So we'll be taking a look
at this at the command line.

112
00:05:31,040 --> 00:05:34,800
Well, that's great, but what
does this actually do for us?

113
00:05:34,800 --> 00:05:36,470
Well, what it's
going to do is it's

114
00:05:36,470 --> 00:05:39,350
going to change
the default values

115
00:05:39,350 --> 00:05:41,580
for these different options.

116
00:05:41,580 --> 00:05:46,220
So first, for
timers throttle spf,

117
00:05:46,220 --> 00:05:50,390
it's going to change the
initial, the secondary,

118
00:05:50,390 --> 00:05:56,284
and the max timers from their
defaults of 5,000, 10,000,

119
00:05:56,284 --> 00:06:00,740
and 10, to be 50, 200, and 5.

120
00:06:00,740 --> 00:06:05,840
For timers throttle lsa all,
once again, the same three

121
00:06:05,840 --> 00:06:06,560
timers.

122
00:06:06,560 --> 00:06:08,990
It has initial,
secondary, and max.

123
00:06:08,990 --> 00:06:14,930
These are going to change from
0, 5000, and 5, to 50, 200,

124
00:06:14,930 --> 00:06:15,890
and 5.

125
00:06:15,890 --> 00:06:18,290
You can see these are
pretty much the same timers

126
00:06:18,290 --> 00:06:21,080
that it uses for
timers throttle spf.

127
00:06:21,080 --> 00:06:23,360
In other words, it's
going to run shortest path

128
00:06:23,360 --> 00:06:26,330
first the same way
it's sending updates.

129
00:06:26,330 --> 00:06:29,870
As far as timers
for lsa arrival,

130
00:06:29,870 --> 00:06:32,210
these don't really
have a category.

131
00:06:32,210 --> 00:06:37,140
It's just it changes the
default from 1,000 to 100.

132
00:06:37,140 --> 00:06:38,610
Now, there's a
lot of information

133
00:06:38,610 --> 00:06:42,030
out there on how you can tweak
these timers to get the best

134
00:06:42,030 --> 00:06:44,820
possible performance
out of OSPF,

135
00:06:44,820 --> 00:06:47,130
but Cisco has done
a lot of research,

136
00:06:47,130 --> 00:06:51,540
and these new optimized values
actually work really, really

137
00:06:51,540 --> 00:06:53,980
well in most environments.

138
00:06:53,980 --> 00:06:55,950
So in most environments,
you're probably

139
00:06:55,950 --> 00:06:57,960
not going to have to
change these settings.

140
00:06:57,960 --> 00:07:00,720
You just want to make sure
that this feature is actually

141
00:07:00,720 --> 00:07:01,810
enabled.

142
00:07:01,810 --> 00:07:05,640
So now what we want to do is
jump into the command line

143
00:07:05,640 --> 00:07:09,750
and just take a look at how do
we change the neighbor timers,

144
00:07:09,750 --> 00:07:11,647
as well as we'll take
a look at this feature,

145
00:07:11,647 --> 00:07:13,230
we'll turn it on and
off, and just see

146
00:07:13,230 --> 00:07:14,680
how it changes the numbers.

147
00:07:14,680 --> 00:07:17,658
So let's start by jumping
over to our topology,

148
00:07:17,658 --> 00:07:19,200
just take a look at
where we're going

149
00:07:19,200 --> 00:07:21,180
to change the neighbor
timers, and then we'll

150
00:07:21,180 --> 00:07:23,440
jump on to the command line.

151
00:07:23,440 --> 00:07:25,020
So for a neighborships,
we're just

152
00:07:25,020 --> 00:07:26,640
going to focus on
this neighborship

153
00:07:26,640 --> 00:07:31,690
up here in area 200 between
router 6 and router 10.

154
00:07:31,690 --> 00:07:33,030
So we'll start with that.

155
00:07:33,030 --> 00:07:36,960
And then again, we'll look at
our routing default optimized

156
00:07:36,960 --> 00:07:37,570
feature.

157
00:07:37,570 --> 00:07:40,640
So let's jump into
the command line.

158
00:07:40,640 --> 00:07:42,370
So we're going to
start here by observing

159
00:07:42,370 --> 00:07:44,830
what happens by default
between router 6

160
00:07:44,830 --> 00:07:47,860
and router 10 when
a failure occurs.

161
00:07:47,860 --> 00:07:49,240
And to see what's
going on, we're

162
00:07:49,240 --> 00:07:51,550
just going to use a split
window here so that we

163
00:07:51,550 --> 00:07:54,280
can see router 6 and
10 at the same time

164
00:07:54,280 --> 00:07:56,690
as we go through
this demonstration.

165
00:07:56,690 --> 00:07:59,440
So let's start by
going to router 10

166
00:07:59,440 --> 00:08:02,380
and shutting down its
interface to router 6,

167
00:08:02,380 --> 00:08:07,210
and see how long it takes router
6 to figure out what happened.

168
00:08:07,210 --> 00:08:10,410
So simply go into configuration
mode for interface gigabit

169
00:08:10,410 --> 00:08:12,730
1, and we will simply do a shut.

170
00:08:12,730 --> 00:08:16,390
Notice router 10 knows
immediately because, of course,

171
00:08:16,390 --> 00:08:18,460
it's a direct link failure.

172
00:08:18,460 --> 00:08:20,820
So router 10, of course,
knows immediately

173
00:08:20,820 --> 00:08:23,220
that that neighborship
is going to go down.

174
00:08:23,220 --> 00:08:26,520
Notice, however, that
router 6 still doesn't know.

175
00:08:26,520 --> 00:08:29,165
Now, this is going
to take 40 seconds,

176
00:08:29,165 --> 00:08:31,290
so we're not going to sit
here necessarily and wait

177
00:08:31,290 --> 00:08:35,100
the whole 40 seconds, but I'll
fast forward until router 6

178
00:08:35,100 --> 00:08:37,120
sees that that
neighborship is down.

179
00:08:37,120 --> 00:08:39,720
So here we can see
router 6 has now realized

180
00:08:39,720 --> 00:08:41,789
that the neighborship
is down, and that

181
00:08:41,789 --> 00:08:44,610
did take right around
40 seconds or so.

182
00:08:44,610 --> 00:08:46,950
So let's go ahead and bring
this connection back up.

183
00:08:46,950 --> 00:08:50,250
We see that both sides now
realize the neighborship is up.

184
00:08:50,250 --> 00:08:53,040
Now, let's adjust
those Hello timers

185
00:08:53,040 --> 00:08:55,660
and see the difference
in performance.

186
00:08:55,660 --> 00:08:59,220
So what we want to do is
we want to say ip ospf,

187
00:08:59,220 --> 00:09:03,510
and we're going to use a feature
called the dead-interval, which

188
00:09:03,510 --> 00:09:08,850
you can set this, of course,
to be from 1 to 65535 seconds.

189
00:09:08,850 --> 00:09:14,160
Or you can go with this
special command, minimal.

190
00:09:14,160 --> 00:09:17,160
If you do minimal, it
will set it to one second

191
00:09:17,160 --> 00:09:18,990
as it says there
on the description.

192
00:09:18,990 --> 00:09:21,810
However, it gives
you this other option

193
00:09:21,810 --> 00:09:24,240
to configure the
Hello multiplier.

194
00:09:24,240 --> 00:09:28,380
If we do the Hello multiplier,
we can set this from 3 to 20.

195
00:09:28,380 --> 00:09:31,560
Now, as we mentioned
in our lecture,

196
00:09:31,560 --> 00:09:35,560
it's really best to keep
this at a 4-to-1 ratio.

197
00:09:35,560 --> 00:09:39,330
So to get four
Hellos in one second,

198
00:09:39,330 --> 00:09:42,130
we would want to send,
well, four of them.

199
00:09:42,130 --> 00:09:46,420
And that means it's going to
send every 250 milliseconds.

200
00:09:46,420 --> 00:09:49,200
Now we, of course, need to set
these timers exactly the same

201
00:09:49,200 --> 00:09:51,770
on router 6.

202
00:09:51,770 --> 00:09:53,435
On router 6, it's gigabit 2.

203
00:09:53,435 --> 00:09:58,380

204
00:09:58,380 --> 00:10:00,520
If we look at our
OSPF interface,

205
00:10:00,520 --> 00:10:02,980
we can see that our Hello
interval is, in fact,

206
00:10:02,980 --> 00:10:07,960
set to 250 milliseconds with
a Dead timer of one second.

207
00:10:07,960 --> 00:10:10,380
So now let's go ahead and
shut down our interface again

208
00:10:10,380 --> 00:10:12,720
on router 10 and see
how long it takes

209
00:10:12,720 --> 00:10:14,580
router 6 to figure it out.

210
00:10:14,580 --> 00:10:20,360

211
00:10:20,360 --> 00:10:24,170
Notice now, it was
within one second, much,

212
00:10:24,170 --> 00:10:28,600
much faster and better
performance using the faster

213
00:10:28,600 --> 00:10:29,440
timers.

214
00:10:29,440 --> 00:10:31,880
So let me take this
at a split window now.

215
00:10:31,880 --> 00:10:34,460
So we just broke router
10 back over here.

216
00:10:34,460 --> 00:10:36,520
So now let's take a
look at those default

217
00:10:36,520 --> 00:10:40,000
timers for OSPF and LSA.

218
00:10:40,000 --> 00:10:43,330
If we just do a show IP
OSPF, we can see here

219
00:10:43,330 --> 00:10:47,480
that the timers are set to
what they were on our table.

220
00:10:47,480 --> 00:10:52,382
We can see here that it's set
to 50, 200, 5,000, 500, 200,

221
00:10:52,382 --> 00:10:54,700
and 5000.

222
00:10:54,700 --> 00:10:58,150
If however, we were to turn off
the routing-default-optimize

223
00:10:58,150 --> 00:11:00,550
feature, which we can
do here simply by saying

224
00:11:00,550 --> 00:11:05,680
no routing-default-optimize
and then do a show IP OSPF,

225
00:11:05,680 --> 00:11:08,410
we can see that these timers
are back to the 5,000,

226
00:11:08,410 --> 00:11:11,450
10,000, 10,000, and so on.

227
00:11:11,450 --> 00:11:14,500
So again, this is a good feature
to probably just leave on.

228
00:11:14,500 --> 00:11:15,940
We'll turn it back on.

229
00:11:15,940 --> 00:11:17,360
Verify it again.

230
00:11:17,360 --> 00:11:19,630
And we can see they're
back to the fast timers.

231
00:11:19,630 --> 00:11:25,090
But this feature can be enabled
or disabled if you want.

232
00:11:25,090 --> 00:11:29,410
In this video, we looked at how
to optimize some of the OSPF

233
00:11:29,410 --> 00:11:34,480
timers, specifically, we looked
at the new default values

234
00:11:34,480 --> 00:11:39,850
for the SPF and the LSA timers
using our routing default

235
00:11:39,850 --> 00:11:41,440
optimize feature.

236
00:11:41,440 --> 00:11:44,740
And we looked at how to optimize
the detection of a neighbor

237
00:11:44,740 --> 00:11:47,080
failure by changing
the Hello and Dead

238
00:11:47,080 --> 00:11:50,440
intervals to their
minimum values of one

239
00:11:50,440 --> 00:11:55,180
second for the Dead, 250
milliseconds for the Hellos.

240
00:11:55,180 --> 00:11:57,250
I hope this has been
informative for you.

241
00:11:57,250 --> 00:12:00,240
And I'd like to thank
you for viewing.

242
00:12:00,240 --> 00:12:04,000