Introduction to Redistribution

(Rene Molenaar) #1

This topic is to discuss the following lesson:

(Sascha D) #2

Hey René,
you missed something in the command syntax. After I have read your lesson, I tried the example. With the shown syntax John still doesn’t advertise the Loopback addresses and the connected network. You probably wrote it in the lesson, but I had to complete:

John(config)#router eigrp 12
John(config-router)#redistribute rip
John(config-router)#redistribute connected

John(config)#router rip
John(config-router)#redistribute eigrp 12
John(config-router)#redistribute connected

With this syntax, John started to advertise the Loopback and his directly connected networks. I hope it’s no mistake from my side.

Have a good one!

(Rene Molenaar) #3

Hi Sascha,

That’s right, this tutorial is just an overview of redistribution but I didn’t show the full config. Here’s the configuration example for those 3 routers:

You don’t have to use “redistribute connected” in this example. The 192.168.12.0 /24 network has been advertised in EIGRP so it will be redistributed into RIP. The 192.168.23.0 /24 network has been advertised in RIP so it will be redistributed into EIGRP. If we don’t use the “network” command for these networks then you’ll have to use redistribute connected.

Oh and also be careful not to forget the seed metric…RIP and EIGRP require it or the metric will be infinite.

Rene

(Diana M) #4

Hello Rene,
I have a question, maybe it’s a little silly, but what happen when both routing protocols (EIGRP, OSPF) have the same IP networks, for instance 10.0.0.0, and you redistribute between protocols?
How the routers can distinguish between one network or another?
Thanks!!

(Rene Molenaar) #5

Hi Diana,

When two routing protocols advertise the same prefix (network + subnet mask) then the administrative distance decides which one we will use:

Rene

(Jason W) #6

Rene in your example of Router Jack, John and James – Redistribution occurs in John. John has both EIGRP and RIP. This single router is running 2 protocols thus redistribution is done on it. What if Router 1 was running RIP only. Router 2 is running EIGRP only. Router 1 is connected to Router 2. R1--------R2. Is that possible? That link connecting R1 to R2 has to be either running EIGRP or RIP? -Thus R1 or R2 has to be running two protocols right?

(Rene Molenaar) #7

Hi Jason,

That’s not going to work, both routers have to use the same protocol on the interface.

The “boundary” between two routing protocols is on a router, not on the link.

Rene

(Chheang C) #8

Hi Rane,

I don’t understand Distance EIGRP (Internal) (External) with my diagram.
When I change distance EIGRP 90 105, the flow from R1-R2-R3-R4.

Best Regards,
Chhayheng

(Andrew P) #9

Chheang,
I am sorry, but there doesn’t appear to be enough information to answer your question. I don’t see a diagram in the lesson that talks about R1 - R4 (or even one that has 4 routers).

Could you provide more details?

(sims) #10

Hi,
What does it mean “Redistribution happens outbound.”
Thanks

(Rene Molenaar) #11

Hi Sims,

This means that when you configure redistribution on a router, it doesn’t change the routing table of router itself. The only thing that changes are the routing updates that the router sends to its neighbors.

Rene

(Prem p) #12

Redistribution also adds another problem. If you “import” routing information from one routing protocol into another it’s possible to create routing loops.
could you pls provide insight about routing loops in case of above sceanrio?

(Lazaros Agapides) #13

Hello Prem.

Although it is possible to create a routing loop in the topology described in Rene’s lesson, let’s look at a simpler scenario to understand how a routing look can occur with redistribution.

Let’s say that we have two different routing protocols running on a network. One region of the network is using OSPF and the other RIP. Let’s say that there are two redistribution points, that is, two connections between the OSPF area and the RIP domain (for lack of a better term). So let’s say we have mutual redistribution (in both directions) at both redistribution points. If you left the network like this, you would create a routing loop, since both redistribution points will have multiple routes for the same networks, and install the OSPF route in the routing table since it has the lower AD. So if RP1 wants to send a packet to the RIP network, it will look in its routing table, and obviously choose the OSPF External route since it should be the only one installed in the routing
table (because OSPF has a lower AD than RIP), and forward it to RP2. RP2 would do the same, and forward it back to RP1, hence creating a routing loop.

In order to resolve such issues, we have to tweak the Administrative Distance in such a way so that loops would be eliminated.

I hope this has been helpful!

Laz

(Mahmoud H) #14

Hi Lazaros,
When we do redistribution from Eigrp to OSPF vice versa, which exactly routes are advertised, the best routes in routing table it self OR all routes of EIGRP (or OSPF) before comparing the Admin Distance between different Protocols in routing table.?

(Lazaros Agapides) #15

Hello Mahmoud

This is a very good question and it shows that you’re thinking analytically about routing and redistribution in particular!

First of all it is important to note that the routing table is itself a compilation of the best route to each destination. The routing table does not have more than one route for each destination unless the metric is exactly the same. So in essence, there is no such thing as “the best routes in the routing table” since by definition, the best route for each destination is listed there.

Secondly, when redistribution between one routing protocol and another occurs, say from EIGRP to OSPF, it occurs within one device, that is, the router which has one interface participating in EIGRP and another interface participating in OSPF. Within this router, EIGRP informs OSPF how network A.B.C.0 can be reached. The OSPF process informs other routers in the OSPF area using the OSPF mechanisms and indicates that this router was learned from an external source with the appropriate code in the routing table. The OSPF routers create their routing tables accordingly.

Now if there is a case where one destination can be reached using two different routing protocols, and thus using two different administrative distances, then the route with the smallest administrative distance will be intalled into the routing table. The other will be rejected.

I hope this has been helpful!

Laz

(Mahmoud H) #16

Lazaros,
Many thanks for your usual support

1 Like
(Aleksandr S) #17

Hello! I have a question. How to need redistribution connected networks in OSPF “best practices”? Across

network X.X.X.X X.X.X.X area 0

or

redistribute connected subnets

(Lazaros Agapides) #18

Hello Aleksandr

If I have understood your question correctly, you are asking what is the best practice to redistribute connected networks in OSPF to other AS’s that are running different routing protocols.

The two commands that you mention in your post provide two different functionalities and are not two different ways of achieving the same thing.

The first command network X.X.X.X. X.X.X.X area 0 is a command that just indicates that you want the specific network, which is directly connected to one of the interfaces of the router, to participate in OSPF. The information concerning this network will be shared within the OSPF AS. Without additional configurations, Information about this network will NOT be shared with other routing protocols in other AS’s.

In order to share this information with routers in other AS’s that are running other routing protocols, it is necessary to indicate this using the redistribute command. Specifically, using the command you stated in your post, the following would occur:

* The router redistributes all connected routes that are subnetted to external AS’s
* The connected keyword refers to routes that are established automatically by virtue of having enabled IP on an interface.

I hope this has been helpful!

Laz

(Aleksandr S) #19

Thanks for the answer. But I had in mind the situation when i redistribute to inter network area0. And I can redistribute whith command

network x.x.x.x x.x.x.x area0

or

redistribute connected subnets

But how more correctly?

(Lazaros Agapides) #20

Hello Aleksandr

The term redistribution is used to describe the mechanism by which network routes are distributed from one dynamic routing protocol to another. Redistribution is not when OSPF routes are learned from one OSPF area to another. Routes from one OSPF area to another are handled by the ABR by using LSA Type 3 packets which are summaries of networks found within each area. These summaries get advertised by default without any additional configuration.

Having said that, again, the two commands you mention about have two different functions.

The first command network X.X.X.X. X.X.X.X area 0 is a command that just indicates that you want the specific network, which is directly connected to one of the interfaces of the router, to participate in OSPF. The information concerning this network will be shared within the OSPF area. This will also automatically be shared with other OSPF areas via the ABRs that exist between OSPF areas using Type 3 LSAs but not with any other routing protocol.

The second command redistributes routes that are directly connected to router to other routing protocols.

Again, specifically using the command you stated in your post, the following would occur:

* The router redistributes all connected routes that are subnetted to external AS’s
* The connected keyword refers to routes that are established automatically by virtue of having enabled IP on an interface.

I hope this has been helpful!

Laz