Dynamic Route Failover

[Jesse Hottle]

Hello,

I am trying to get our network traffic to dynamically pick the best route out/in in case of a link failure. I have attached a diagram showing our current router links. The majority of the links are connected to the end-point routers using wireless backhaul bridges. We have two connections forming a ring back to our central noc. (DS3, and PTP fiber connections). Can any one help with this?

[Jesse Hottle]

Hello Istvan,

On the routers running ospf do I still need to add a default route on each router?

[Istvan_Rabai]

Hi Jesse,

By the suggested config Router A should inject a default route into the internal ospf domain.

This is why I put the "default-information originate" command into the ospf config of router A.

So no other default route config is necessary.

Cheers:

Istvan

[Jesse Hottle]

Thanks for the response..

Also, on all the router interfaces that are directly connected to other routers, should I always use the ip ospf network point-to-point command on that specifc interfaces?

Also, on the ospf network configs you have an inteface assigned as 192.168.1.1 and 192.168.1.5, but within the global ospf config you list the adversitzed networks as 192.168.1.1 0.0.0.0 and 192.168.1.5 0.0.0.0 (why are these not 192.168.1.0 0.0.0.3)..

And on the interface that is connection to a switch for end-users you have 192.168.2.1 255.255.255.0, but on the ospf global you have 192.168.2.0 0.0.0.255)?

Just trying to clarify this up some for my own knowledge.

Thanks again,

J

[Istvan_Rabai]

Hi Jesse,

Your 1st question:

I used to specifiy "ip ospf network point-to-point" on links that are point-to-point.

The link between Router A and Router F is a fiber optic circuit and it is point-to-point (only these 2 routers are on this link).

If you specify the network type for ospf "ip ospf network point-to-point", the 2 routers form a neighbor relationship automatically but without the use of a designated router (DR)and backup designated router (BDR).

DR and BDR are necessary only when you have more than 2 routers on a broadcast link, or a full mesh of connections on a non-broadcast multiaccess link (e.g. frame relay).

Not using a DR and BDR saves memory and processor cycles for the routers.

Your 2nd question:

0.0.0.0 wildcard mask specifies that all bits of the given address need to be compared with the interface addresses.

The implication of this is that only one interface will be included in the ospf process: the one that has the exact 192.168.1.1 address.

192.168.1.0 0.0.0.3 will achieve the same result in this specific configuration, so you can use it as well.

Your 3rd question:

192.168.2.1 255.255.255.0 specifies the ip address of an interface and its subnet mask.

192.168.2.0 0.0.0.255 uses a wildcard mask instead of a subnet mask.

This wildcard mask means this: 00000000.00000000.00000000.11111111

The bits of the address 192.168.2.0 must be compared with the addresses of the interfaces on the router, where 0s are specified in the wildcard mask.

Where 1s are specified, these are "don't care" bits, there is no need for comparison.

The above wildcard means then: only the leftmost 3 octets must be compared. If these bits match, then the interface is selected under the ospf process.

This is not the same as the subnet mask, where the 1s and 0s must be contiguous.

For a wildcard mask there is not a requirement for 0s and 1s to be contiguous You can have a wildcard mask like this:

00000000.00000000.11111101.11111111

(0.0.253.255)

In this case bit 2 of the 3rd octet is 0, all other bits are 1.

This requires the respective bit of that octet to be compared with the interface addresses. All bits with value 1 don't have to be compared. Bit 1s are called "don't care" bits.

Cheers:

Istvan

[Istvan_Rabai]

Hi Jesse,

Although you didn't mention if you will use a private address space within your internal ospf domain, my supposition is that you will use a private address space.

If this is true, then let me make an addition to the configurations I posted for you beforehand:

If you use private addresses in your internal ospf domain, you will need to do NAT address translation. It is configured on Router A the following way ( I post the modified config for Router A here):

[Router A config, sample values]

interface serial 1/0

description DS3 to Router B 7206

ip address 192.168.1.1 255.255.255.252

ip nat inside [part of the NAT config]

interface fastethernet 1/0

description 100Mbps fiber to Router F 7206

ip address 192.168.1.5 255.255.255.252

ip ospf network point-to-point

ip nat inside [part of the NAT config]

interface fastethernet 1/1

description 100Mbps to ISP

ip address x.x.x.1 x.x.x.x

ip nat outside [part of the NAT config]

no cdp enable [this may be needed for security purposes, the router will not send CDP packets to the ISP]

router ospf 1

log-adjacency-changes

network 192.168.1.1 0.0.0.0 area 0 [this places interface serial 1/0 into area 0]

network 192.168.1.5 0.0.0.0 area 0 [this places interface fastethernet 1/0 into area 0]

default-information originate [this command causes ospf to send a default route to the internal ospf network, so internal routers know how to reach the outside world]

ip route 0.0.0.0 0.0.0.0 x.x.x.2 [default route to the ISP]

ip nat inside source list 1 interface fastethernet1/1 overload [part of the NAT config]

access-list 1 permit 192.168.0.0 0.0.255.255 [part of the NAT config]

I hope this will give you an additional help in your configuration work.

Cheers:

Istvan

[Jesse Hottle]

Hello,

I had not planned on using NAT, as I am afraid this will causes issues when we begin using BGP along with OSPF for other networks we are going to be routing through our network backhauls. Woudl this cause a problem? Why would you recommend NAT?

Thanks again for the responses.

J

[Istvan_Rabai]

Hi Jesse,

NAT is only needed when you use a private address range within your internal ospf domain. Private addresses are not routable on the Internet.

If you have enough public ip addresses for use within your network, then it is not necessary to use NAT.

Cheers:

Istvan

[Jesse Hottle]

Hello again,

I got a question on Secondary router interfaces. Do I need to use any ip ospf commands on these secondary interfaces (i.e f0/0.1 and f0/0.2)? Or can I just enable OSPF on the primary interface of f0/0?

Thanks,

Jesse

[Istvan_Rabai]

Hi Jesse,

What what would be the purpose of using subinterfaces?

Generally speaking, there is no problem of using them, but why isn't the main interface enough?

Cheers:

Istvan

[Jesse Hottle]

Hello,

On one of the router interfaces for client nodes, switches, etc.. I have the interface connected to a switch trunk port. One the switch have I have other devices on seperate vlans. I need each vlan to be on its own subnet, hence the sub-interfaces on the router side interface.

J

[Istvan_Rabai]

I undestand.

This is a an intervlan routing configuration then.

Its popular name is router on a stick.

If you need help in this config, please tell me.

How is the ospf configuration proceeding. Is it working well?

Cheers:

Istvan

[Jesse Hottle]

Hello,

I got the router on a stick config covered, just wondering how this would work with OSPF. I will be integrating this setup into our network this evening and will let you know how things go.

thanks again.

j

[Jesse Hottle]

Hello,

I got all the routers installed and running OSPF. Dynamic fallover is working well, however there is an issue with the paths traffic is following. Traffic seems to be unbalanced between different outbound routes. Is there a way for specify a prefered route for each interface or subnet?

Thanks again,

j

[Jesse Hottle]

Also, I got one L3 3550 running three ospf interfaces. Two of the interfaces are connected to seperate links on the ring, but the third is attached to a single router. I am not sure if I have configured this router correctly.

thanks,

j

[Istvan_Rabai]

For the L3 3550 you can post interface and ospf configs here and I will check.

Cheers:

Istvan