|
0:00:15
|
OK. So our next task 16.3 for Load Balancing Calls
|
|
0:00:20
|
We´re told to provision a method so that any
|
|
0:00:22
|
call place from Branch2 phones to the phones that Corporate
|
|
0:00:26
|
headquarter or Branch1 sites are sent to
|
|
0:00:29
|
both the subscriber at 177.1.10.20
|
|
0:00:33
|
and the publisher at .10 servers in a load balance fashion
|
|
0:00:38
|
so that neither gets overwhelm with too many calls.
|
|
0:00:42
|
OK? Well we had a raise our dial peer from earlier
|
|
0:00:47
|
that I´ll just go ahead and copy and paste back in here
|
|
0:00:50
|
since we had essentially gotten rid of it.
|
|
0:00:54
|
And I believe that one was pointing to, don´t recall if it´s
|
|
0:00:58
|
pointing to the publisher or subscriber, but we´ll put in one for each
|
|
0:01:02
|
and the reason we raised it is so that they would be in the proper order.
|
|
0:01:06
|
So we´re going to dial peer voice 10 VOIP
|
|
0:01:09
|
destination pattern, left bracket, right bracket
|
|
0:01:13
|
and then this can either be a dash if there´s a range of numbers
|
|
0:01:17
|
or even if they are numbers that are not contiguous to
|
|
0:01:20
|
one another, 1 and 2 are, but they could be 18
|
|
0:01:23
|
if there´s no dash, then it´s not a range. It´s saying either
|
|
0:01:27
|
1 or anyone of these digits if we had maybe
|
|
0:01:30
|
5 digits in here with no dashes, anyone of these digits
|
|
0:01:33
|
can successfully be match for this one position, but only one digit.
|
|
0:01:38
|
So either 1 or 2..., so 1000 series,
|
|
0:01:43
|
2000 series are sent via SIP, there´s already a SIP
|
|
0:01:46
|
trunk back at call manager and this one sending to the publisher
|
|
0:01:51
|
dtmf relay is rtp nte, rfc 2833 which is what we´re using on our
|
|
0:01:55
|
SIP phones and what´s preferred over a SIP trunk
|
|
0:02:00
|
kpml being the back up and.
|
|
0:02:05
|
We were told to load balance these calls,
|
|
0:02:07
|
so we´re going to add another dial peer here.
|
|
0:02:11
|
For dial peer 11 pointing to same destination
|
|
0:02:16
|
pattern but .20 and we´re not going to add preference.
|
|
0:02:21
|
Now, no hunt stop or don´t stop hunting
|
|
0:02:26
|
or allow hunting is the default for dial peers.
|
|
0:02:32
|
At least the default for dial peers is you create from the command line
|
|
0:02:36
|
dial peers that are created for you from things like
|
|
0:02:40
|
ephone and ephone DN, the pots dial peers that are created there
|
|
0:02:46
|
hunts stop is actually the default built in to those templates. So
|
|
0:02:50
|
stop hunting, so if we want those to hunt we would have
|
|
0:02:52
|
explicitly say no hunt stop, but here if we say no hunt stop.
|
|
0:02:58
|
And we do show run type2 begin with dial dash peer.
|
|
0:03:04
|
We see that nothing shows up there, that´s because it´s the default,
|
|
0:03:08
|
don´t stop hunting or allow hunting. The problem is
|
|
0:03:11
|
there´s no preference, which preference 0
|
|
0:03:14
|
sorry for the scroll, preference 0 is the default, but
|
|
0:03:19
|
they´re both preference 0, so which one is going to take it?
|
|
0:03:23
|
Well, by default everything is random
|
|
0:03:26
|
however, there is more to dial peers than
|
|
0:03:31
|
meets the eye and certainly the behaviour can be different
|
|
0:03:34
|
from that from CUCM. So we´ve got dial peer
|
|
0:03:37
|
hunt, now this is not a hunt group, we´ll tak about that in,
|
|
0:03:42
|
I believe, not the next module but this is module 16
|
|
0:03:47
|
I belive in module 18 we deal with all the call coverage
|
|
0:03:50
|
and DACD and hunt groups in IOS and CME.
|
|
0:03:54
|
But here dial peer hunt refers to the hunting algorithm
|
|
0:03:59
|
and 0 is the default, so this is a global config command
|
|
0:04:03
|
notice I´m not inside the dial peer anymore.
|
|
0:04:05
|
If I say dial peer hunt 0 and I say do show run type2 section
|
|
0:04:10
|
hunt or I should say maybe type2 include hunt.
|
|
0:04:15
|
It doesn´t show up anywhere, if I change
|
|
0:04:18
|
dial peer hunt to let´s say 7,
|
|
0:04:22
|
and do that show, it does show up
|
|
0:04:25
|
So we know 0 is the default and by looking
|
|
0:04:27
|
at the command reference, we could know that 0 is the default.
|
|
0:04:30
|
The default is to first try to match the longest phone number,
|
|
0:04:34
|
then if that´s a tie, the explicit preference
|
|
0:04:38
|
then if that´s a tie, random selection
|
|
0:04:42
|
I might say good so it´s going to choose both of these at some point.
|
|
0:04:45
|
Yes, it will but it will not be in a round robin
|
|
0:04:48
|
load balance fashion, it coule be
|
|
0:04:51
|
choose dial peer 10, dial peer 11,
|
|
0:04:54
|
dial peer 10, 10, 10, 10,10,10,10,10,10,11.
|
|
0:04:58
|
10, 10, 10, 11, 11, 11, 11, 10, 11, 11, 11, 11
|
|
0:05:02
|
and so on, it´s random. So anyone has a chance of being the winner.
|
|
0:05:08
|
It will not load balane, let´s look at some of the others.
|
|
0:05:11
|
We´ll start at the bottom, least recent use. Well that´s not bad
|
|
0:05:15
|
but if the only dtermining factor
|
|
0:05:18
|
or completely random, now it still has to match but
|
|
0:05:20
|
completely random, I could have 9..
|
|
0:05:26
|
911 and 91. each with different preferences
|
|
0:05:32
|
and neither the preferences nor the longest match has any
|
|
0:05:36
|
choice, it just has to match and it´ll match randomly or choose randomly.
|
|
0:05:41
|
Anyway, you can read on all these, but the one we are probably
|
|
0:05:44
|
going to want is 1 which is the same as 0,
|
|
0:05:46
|
longest match in phone number then explicit preference
|
|
0:05:50
|
then if neither of those 2 tie out, then you
|
|
0:05:54
|
change from 0 is least recent use.
|
|
0:05:58
|
So let´s change this to one and now are dial peers
|
|
0:06:04
|
will go back and forth as to which one is chosen.
|
|
0:06:08
|
So to give an example, see what debugs we have turned on.
|
|
0:06:13
|
Alright, let´s do a call over to 1001 and
|
|
0:06:20
|
we´re going to answer it, go to mute.
|
|
0:06:23
|
So we don´t have too much feedback and do a show
|
|
0:06:26
|
call active voice compact.
|
|
0:06:32
|
And that´s actually not showing which
|
|
0:06:36
|
dial peer we´re going out of, so let´s do show voice.
|
|
0:06:40
|
Sorry, show call active voice type 2 include here ID
|
|
0:06:46
|
and this shows us that we came inbound through 20001
|
|
0:06:51
|
which is a POT style up here from skinny ephone
|
|
0:06:55
|
and outbound through dial peer 10
|
|
0:06:59
|
OK? tear down that call, there´s nothing active
|
|
0:07:04
|
do the call again, just hit redial from the same phone to the same phone
|
|
0:07:12
|
And this time we chose outbound peer 11. OK, tear down the call.
|
|
0:07:20
|
Do redial again.
|
|
0:07:26
|
OK, this time we chose peer 10 and once again
|
|
0:07:34
|
redial from the same phone to the same phone.
|
|
0:07:39
|
We chose peer 11 as we can see, so it is load balancing.
|
|
0:07:44
|
The reason I don´t typically like to do the
|
|
0:07:46
|
sample demonstration with the random
|
|
0:07:49
|
is because sometimes it will actually load balance but sometimes
|
|
0:07:52
|
it will choose one for like 10 years, not really
|
|
0:07:59
|
but I mean if I´m exaggerating, but it will choose it for like
|
|
0:08:02
|
it seems like over a minute of me just continuing to do the redial and
|
|
0:08:06
|
check the config before I actually hit the other one
|
|
0:08:09
|
so it truly is random and sometimes it´s hard to actually produce
|
|
0:08:13
|
the good example of it being completely random,
|
|
0:08:16
|
we can try it here real quick, if it doesn´t
|
|
0:08:18
|
happen quickly, I´m going to change it back. Alright.
|
|
0:08:24
|
OK, so we currently have nothing let´s do a undebug all
|
|
0:08:29
|
we´ve already seen all the voice translations,
|
|
0:08:31
|
let´s clear off the screen, alright. So redial
|
|
0:08:39
|
There´s peer 10, we´ll do a redial
|
|
0:08:44
|
There´s peer 10, OK. So let us chosen the same one in a row
|
|
0:08:48
|
couple times. Now it shows peer 11.
|
|
0:08:53
|
Great and it looks like it´s working actually pretty well this time.
|
|
0:08:57
|
again it´s hard to predict, no it shows 11
|
|
0:09:04
|
and it still shows 11, OK. So
|
|
0:09:10
|
still chose 11, try it again.
|
|
0:09:17
|
still choosing 11. OK, So we can clearly see that it
|
|
0:09:20
|
chose 10 a few times, but now it´s almost always choosing 11
|
|
0:09:24
|
The poing is it´s certainly not load balancing them, we go back
|
|
0:09:29
|
and then we´ll just do it a couple times.
|
|
0:09:37
|
Call is up peer 1o, redial
|
|
0:09:43
|
peer 11 and one more time just to show that it´s back to peer 10
|
|
0:09:46
|
guaranteed, back to peer 10. OK
|
|
0:09:51
|
So we are in fact load balance.
|
|
0:09:54
|
We´ll write this router config and clear off our screen
|
|
0:09:58
|
and we´ve taken care of that task.
|
|
0:10:00
|
|
