|
0:00:13
|
The next network type we have is point-to-multipoint non-broadcast,
|
|
0:00:18
|
which is identical in its operation to point-to-multipoint
|
|
0:00:22
|
with the one exception that we are sending our updates as unicasts as opposed to multicasts.
|
|
0:00:29
|
So, the non-broadcast and point-to-multipoint non-broadcast,
|
|
0:00:33
|
those are using unicast updates.
|
|
0:00:36
|
Broadcast, point-to-point, and point-to-multipoint are all using multicasts.
|
|
0:00:41
|
The multicast difference between them is that broadcast is usually two different addresses.
|
|
0:00:46
|
224.0.0.5 and 224.0.0.6
|
|
0:00:50
|
Where one of them is for the designated router. One of them is for all of the other routers.
|
|
0:00:55
|
The 225.0.0.6, that's for traffic going to the DR.
|
|
0:01:00
|
Point-to-point and point-to-multipoint, they're using just 224.0.0.5,
|
|
0:01:04
|
which is all OSPF routers.
|
|
0:01:08
|
Now, for point-to-multipoint non-broadcast, the case that we would want to use this
|
|
0:01:13
|
is that if we have a partial mesh of Layer 2 circuits
|
|
0:01:18
|
but the underlying netowrk is made up of different bandwidth values on the circuits.
|
|
0:01:23
|
The case that this is typical is that either in a frame-relay or ATM design
|
|
0:01:29
|
where we have router 5 connecting with different virtual circuits to the spokes.
|
|
0:01:36
|
But the virtual circuits are provisioned at different bandwidth levels.
|
|
0:01:40
|
So, let's say for example that router 5's physical interface here is a T1.
|
|
0:01:46
|
T1 is made up of 24 channels of DS0, which are 64 kilobits per second.
|
|
0:01:55
|
So, if we look at the effective bandwidth of this,
|
|
0:01:59
|
24 times 64 kilobits per second, it's 1,536 kilobits per second.
|
|
0:02:05
|
So, actual effective bandwidth value.
|
|
0:02:07
|
There's gonna be some other band signaling for this, which is the additional 8 kilobits per second.
|
|
0:02:15
|
That's not actual effective bandwidth.
|
|
0:02:17
|
So, sometimes when you see this listed as 1544, that's including the signaling.
|
|
0:02:25
|
but the actual useable bandwidth on the DS0 circuit is 64K.
|
|
0:02:29
|
So, let's say that this...
|
|
0:02:31
|
T1 interface, 1536 is broken down into 4 equal assignments.
|
|
0:02:39
|
So, each of this circuits assigned 384 kilobits per second.
|
|
0:02:47
|
So, 504 is 384K.
|
|
0:02:53
|
501, 503, and 502.
|
|
0:02:57
|
Okay, they're all the same bandwidth value.
|
|
0:02:59
|
In this case, it doesn't matter that when router 5 does it's calculation for the path selection
|
|
0:03:05
|
that it uses the single bandwidth value that is applied to the intreface level.
|
|
0:03:13
|
bewcause when we look at router 5's configuration,
|
|
0:03:16
|
if we Show Route Interface Serial 0/0/0,
|
|
0:03:22
|
all of this 4 circuits, they're assigned to the same interface.
|
|
0:03:27
|
This meaens if we Show IP OSPF Interface Serial 0/0/0,
|
|
0:03:32
|
all of these circuits are getting the same underlying cost value.
|
|
0:03:38
|
This is based on the interface's bandwidth value.
|
|
0:03:40
|
If I Show Interface Serial 0/0/0,
|
|
0:03:44
|
it says, "The bandwidth value is 1544."
|
|
0:03:48
|
But let's say that this is different. It's gonna represent the individual circuits.
|
|
0:03:52
|
If the bandwidth were 384 kilobits per second,
|
|
0:03:57
|
when we Show IP OSPF Interface Serial 0/0/0,
|
|
0:04:02
|
and let's just look at the cost,
|
|
0:04:05
|
it says now, "The cost is 260."
|
|
0:04:08
|
So, a 384 kilobit per second circuit by default results in a cost of 260.
|
|
0:04:15
|
But now, let's say that one of these links is not provisioned at that.
|
|
0:04:20
|
Maybe one of them is half of this. They're at 192.
|
|
0:04:24
|
And this link that's provisioned at 192, we'll say that this is router 2's circuit.
|
|
0:04:30
|
192 kilobits per second.
|
|
0:04:33
|
The problem now is that when router 5 looks at a route being learned from 2.
|
|
0:04:40
|
And a route being learned from 3.
|
|
0:04:43
|
They're both gonna get the underlying cost value that are applied to the same main interface.
|
|
0:04:51
|
So, in the case where the cost of the main interface does not reflect the underlying Layer 2 circuits,
|
|
0:04:58
|
that's what point-to-multipoint no-broadcast is for.
|
|
0:05:02
|
So, router 5 wwill still treat this as a collection of point-to-point links.
|
|
0:05:07
|
But we can apply on a per-neighbor basis what is the cost value
|
|
0:05:12
|
for link state updates that we are receiving for that neighbor.
|
|
0:05:15
|
So, now, let's look at what would happen if the bandwidth were 192.
|
|
0:05:20
|
We should see that there cost is gonna be doubled, it's 560.
|
|
0:05:25
|
So, at router 5's linklevel, I could say the IP OSPF network type is point-to-multipoint.
|
|
0:05:32
|
But the non-broadcast version.
|
|
0:05:35
|
Now, under the OSPF process, I could say that for the neighbor 150.42.0.1,
|
|
0:05:43
|
I wanna send unicast to them but I also wanna define their cost.
|
|
0:05:47
|
And theie cost will be 260.
|
|
0:05:51
|
Router 3 and 4, they likewise have the same identical cost.
|
|
0:05:59
|
But for router 2,
|
|
0:06:02
|
this cost is doubeld because the circuit speed is half. So, it's 520.
|
|
0:06:07
|
When we look at the end result to this and Show IP OSPF neighbors,
|
|
0:06:12
|
we see that we still have the adjacencies with all 4 of them.
|
|
0:06:16
|
If we look at the Debug IP Packet Detail,
|
|
0:06:20
|
we would see that the actual transport
|
|
0:06:23
|
of the packets between the neighbors are going to use the unicasts now.
|
|
0:06:29
|
No longer the multicast.
|
|
0:06:36
|
So, we see the unicast packets between the neighbors.
|
|
0:06:41
|
We can also see this is OSPF because it's protocol number 89.
|
|
0:06:47
|
If we look at the Show IP Route OSPF,
|
|
0:06:56
|
we'll see that there's a difference in the cost of the
|
|
0:07:00
|
routes to router 2 versus the other neighbors.
|
|
0:07:04
|
And I believe, let's see, let's look at the Show IP OSPF Interface Serial 0/0/0.
|
|
0:07:11
|
The timers here may be different.
|
|
0:07:16
|
Then the default ones. Let's go to the other side, let's see what the router 1 and 2 have.
|
|
0:07:21
|
Let's Show IP...
|
|
0:07:24
|
OSPF Interface S0/0.1
|
|
0:07:29
|
These are 30 and 120.
|
|
0:07:32
|
Let's see what router 3 has. Show IP OSPF Interface Serial 1/0.
|
|
0:07:41
|
This is 30 and 120.
|
|
0:07:42
|
So point-to-multipoint, point-to-multipoint non-broadcast, they both have the same timers, which is fine.
|
|
0:07:47
|
So, on router 5, let's look again at the...
|
|
0:07:52
|
Actually, what I'll do, let me shut the link down to router 4.
|
|
0:07:57
|
Because that is being preferred now based on the higher bandwidth there.
|
|
0:08:04
|
Now, let's Show IP Route OSPF.
|
|
0:08:11
|
We should see now that there's gonna be a difference in the cost,
|
|
0:08:15
|
between the loopback of router 3 and the loopback of router 2.
|
|
0:08:21
|
Because they're defined different cost values under the process.
|
|
0:08:27
|
Now, at the physical interface, I may need to remove the bandwidth statement here.
|
|
0:08:36
|
But if we look at the configuration under the process,
|
|
0:08:41
|
we see, we have the neighbor statements plus we have the cost value associated with them.
|
|
0:08:46
|
So, if I clear the routing table and then Show IP Route OSPF,
|
|
0:08:55
|
we should see that if we look at just those host routes,
|
|
0:09:06
|
the cost is gonna be different. I may need to clear the process now.
|
|
0:09:09
|
Let's say Clear IP OSPF Process.
|
|
0:09:31
|
So, it's showing 65 for all of these which means the interface cost is overriding this.
|
|
0:09:37
|
Let's say Show IP OSPF Interface Serial 0/0/0.
|
|
0:09:45
|
The cost of the link is 64.
|
|
0:09:49
|
I'm wondering if we need to set this higher. Let's say that...
|
|
0:09:53
|
the cost value is IP OSPF Cost...
|
|
0:10:03
|
we should see that the neighbor's cost is overriding what's on the interface.
|
|
0:10:12
|
So, what I may need to do here is also do this on the remote side, on the other spokes.
|
|
0:10:18
|
But if we look at the... Let's take a look at the documentation here. If we go to the main...
|
|
0:10:26
|
website - the main documentation's site,
|
|
0:10:28
|
this would be under products, IOS
|
|
0:10:32
|
Cisco IOS, 12.4, 12.4T
|
|
0:10:38
|
Then I want the configuration guide for OSPF and also the command reference.
|
|
0:10:43
|
So, on the reference guides, command reference,
|
|
0:10:49
|
under IP Routing, OSPF commands,
|
|
0:10:53
|
then let's say under...
|
|
0:10:59
|
neighbor OSPF...
|
|
0:11:14
|
It says, "To configure OSPF Routers interconnecting with non-broadcast networks,
|
|
0:11:17
|
chooes the neighbor command...
|
|
0:11:21
|
Optionally, assigns a cost to the neighbor in the integer 1 to 65535.
|
|
0:11:27
|
Neighbors with no specific cost configure will assume the cost of the interface
|
|
0:11:31
|
based on the IP OSPF cost command.
|
|
0:11:33
|
For point-to-multipoint interfaces, the cost keyword and the number arguement
|
|
0:11:38
|
are the only option that are applicable. The keyword does not apply to non-broadcast multiaccess networks.
|
|
0:11:43
|
Which would mean, the network type non-broadcast.
|
|
0:11:46
|
So, I'm wondering if the case that 5 has both commands under the routing process.
|
|
0:11:57
|
Let's see if I can remove the first one.
|
|
0:12:02
|
or actually, let's just start this over. Let's say "No router OSPF 1."
|
|
0:12:06
|
So, what I want is just the network statement.
|
|
0:12:13
|
And then theneighbor costs.
|
|
0:12:19
|
So, now, let's say, "Show Rune Section Router OSPF"
|
|
0:12:24
|
Now, it's showing what we just want.
|
|
0:12:25
|
So, I think it's because before we were using netowrk type non-broadcasts,
|
|
0:12:31
|
then I was entering the network command again or the neighbor command again.
|
|
0:12:36
|
So, most likely the first one was overriding it.
|
|
0:12:39
|
So, let's see what's the result in the routing table.
|
|
0:12:44
|
And then I will remove at the link level. So, no IP OSPF Cost.
|
|
0:13:33
|
So, this version may not actually support this then.
|
|
0:13:38
|
We should see a higher cost value that's beign received from 2's loopback as opposed to the other ones.
|
|
0:13:45
|
We could try this on the Ethernet to see whether it was supported then.
|
|
0:13:53
|
But this originally was supposed to be for the case where the PVC speeds are mismatching.
|
|
0:13:59
|
Let's try this one more thing, let's clear the OSPF process.
|
|
0:14:09
|
And see if that affects anything.
|
|
0:14:37
|
So, here, they're taking the cost of 64 form the link level instead of the process level.
|
|
0:14:43
|
Where you could also use this...
|
|
0:14:50
|
There's a question, "What is the network type on router 5's serial?"
|
|
0:14:52
|
It shold be point-to-multipoint non-broadcast.
|
|
0:14:58
|
Let's try it with this, let's set it just to...
|
|
0:15:09
|
Let's set it just to point-to-multipoint and not point-to-multipoint non-broadcast.
|
|
0:15:26
|
So, it's still taking the cost from the link level.
|
|
0:15:30
|
The other case where this would be useful is if the routers are connected over a LAN.
|
|
0:15:37
|
But the physical speeds of the Ethernets are different that are going to the switch.
|
|
0:15:44
|
So, let's say we have a Layer 2 switch in the middle here.
|
|
0:15:47
|
Router 1 is connecting with Fast Ethernet.
|
|
0:15:50
|
Router 3 is connecting with Fast Ethernet.
|
|
0:15:52
|
And router 1 is connecting with Regular Ethernet, so 10 megabits per second.
|
|
0:15:58
|
On router 2 and 3, if this was network-type broadcasts,
|
|
0:16:04
|
they would not be able to tell the difference between the 100 megabits per second connection and the 10.
|
|
0:16:14
|
So, we could change the network-type point-to-multipoint
|
|
0:16:18
|
and then specify on the per-neighbor basis, what the cost value is.
|
|
0:16:23
|
So, it may be that in the newer IOS, this is not supported for the frame-relay interfaces.
|
|
0:16:29
|
But this was the original usage for it.
|
|
0:16:31
|
That if you're PVC speeds are different, you could set the cost on a per-neighbor basis.
|
|
0:16:38
|
But again, the easier design solution for this is to never use a case like this on router 5 to begin with.
|
|
0:16:48
|
So, we'll see as we get into more details with the Layer 3 protocols.
|
|
0:16:52
|
The routing process, the muticast process, the QoS process.
|
|
0:16:56
|
A lot of things do not relate well to configurations
|
|
0:16:59
|
where the Layer 2 network does not map to the Layer 3 network on a one-to-one basis.
|
|
0:17:06
|
So, if router 5 had four separate point-to-point sub-interfaces,
|
|
0:17:11
|
then OSPF would be able to say,
|
|
0:17:13
|
"This subinterfaces have different cost value associated with than the other ones."
|
|
0:17:19
|
Then router 5 would be able to tell the difference between the PVC speeds.
|