Route Redistribution and route maps

[mandeep62]

Hi all,

I'm trying to learn route redistribution. Below I have attached the topology where the network consists of EIGRP and OSPF. I understand that route redistribution is used where we are running multiple routing protocols and we need to translate one into other so that it can be understood by the other. In this topology, R2 and R4 are the source of redistribution as they are edge routers who speak both EIGRP and OSPF. R1 is running EIGRP and R3 is running OSPF.

Firstly, I'm confused about the seed metric. What is seed metric and default-metric? I know that If I want to redistribute EIGRP routes into ospf, I have to translate it by defining the EIGRP metric under ospf. What values should I give under metric? Is it always 10000 1000 255 1 1500 or is it just a random value? If it is a random value, what should be the correct values of EIGRP components given?

Similarly, 100 subnets is specified because that will allow all individual subnets to be sent instead of a summarized route? Is it right?

e.g

router eigrp 100
 network 10.1.1.0 0.0.0.255
 redistribute ospf 1 metric 10000 1000 255 1 1500 route-map ospf>EIGRP

router ospf 1
 redistribute eigrp 100 subnets route-map EIGRP>OSPF
 network 2.2.2.2 0.0.0.0 area 0
 network 10.1.2.0 0.0.0.255 area 0

Now, I have read that mutual redistribution can cause routing loop because routes can be exchanged back and forth between multiple source routing protocols. So, we use route maps.

Can anyone explain how route maps works? Below route-maps are configured on both R2 and R4. I would appreciate if someone could tell how it is filtering the traffic?

route-map ospf>EIGRP deny 5
 match tag 20
!
route-map ospf>EIGRP permit 10
 set tag 10
!
route-map EIGRP>OSPF deny 5
 match tag 10
!
route-map EIGRP>OSPF permit 10
 set tag 20
!

Thanks.

[Seb Rupik]

Hi there,

Each routing protocol has a seed metric (ie default), in the case of EIGRP if not defied gives an infinite metric value. You therefore have two options, either define default-metric under the EIGRP process, or specify a EIGRP metric via the redistribute command as you have done.

Regarding the values, understanding what they mean will help:

Metric	Value
bandwidth	In units of kilobits per second; 10000 for Ethernet
delay	In units of tens of microseconds; for Ethernet it is100 x 10 microseconds = 1 ms
reliability	255 for 100 percent reliability
load	Effective load on the link expressed as a number from 0 to 255 (255 is 100 percent loading)
MTU	Minimum MTU of the path; usually equals that for the Ethernet interface, which is 1500 bytes

These are then ‘tuned’ by the EIGRP K-values….but at this stage just leave those as default!

You assertion about the subnets command is correct.

Regarding using route-maps to prevent loops, pay close attention to the deny statements. If we look at the route-map ’ospf>EIGRP’:

route-map ospf>EIGRP deny 5
 match tag 20
!
route-map ospf>EIGRP permit 10
 set tag 10
!

…the first step says to drop any routes that have a tag of 20, and for all other routes give a tag of 10. Now when EIGRP uses the route-map ‘EIGRP>OSPF’ when the first line is evaluated any routes with a tag of 10 (the ones from the OSPF process) will be dropped, and the rest tagged with a value of 20….which is then used I the previous route-map. So this stops a routers ‘own’ routes being redistributed back to it by a peer router.

Cheers,

Seb.

https://www.cisco.com/c/en/us/support/docs/ip/enhanced-interior-gateway-routing-protocol-eigrp/16406-eigrp-toc.html

https://www.cisco.com/c/en/us/support/docs/ip/enhanced-interior-gateway-routing-protocol-eigrp/8606-redist.html

[mmanthe]

I thought I'd add a couple of points here to help clarify some things and hopefully answer some of your questions as well...

"What is seed metric and default-metric? I know that If I want to redistribute EIGRP routes into ospf, I have to translate it by defining the EIGRP metric under ospf"

This is actually the opposite. You would define EIGRP metrics for OSPF routes that you are distributing *INTO* EIGRP. 

As OSPF and EIGRP are completely different protocols with completely different systems of metrics and calculations - the originating metrics from a source protocol are stripped from routes as they are distributed into a destination protocol - therefor it is sometimes necessary to provide them manually. In the case of routes going INTO EIGRP, you need to define a 'seed metric' to give the incoming routes a baseline of metrics that EIGRP can understand. In the case of OSPF, you *can* manually define a metric if you want - but OSPF can also apply a 'default metric' to incoming, redistributed routes if none are defined manually.

"Similarly, 100 subnets is specified because that will allow all individual subnets to be sent instead of a summarized route? Is it right?"

Essentially, yes. When OSPF was first released it was a 'classful' protocol. It did not 'understand' VLSM or CIDR. The 'subnets' command was added later to allow OSPF to recognize classless subnets.

Now if your question meant to single out the number '100' as some sort of subnet count - then no. That '100' refers to the EIGRP process number, e.g.

router eigrp 100

Since it is possible to run more than one EIGRP (or OSPF, etc.) process on a device, it is necessary to distinguish which process ID you wish to distribute routes from.

Hopefully that adds a little more clarity. 

[mmanthe]

I thought I'd add a couple of points here to help clarify some things and hopefully answer some of your questions as well...

"What is seed metric and default-metric? I know that If I want to redistribute EIGRP routes into ospf, I have to translate it by defining the EIGRP metric under ospf"

This is actually the opposite. You would define EIGRP metrics for OSPF routes that you are distributing *INTO* EIGRP.

As OSPF and EIGRP are completely different protocols with completely different systems of metrics and calculations - the originating metrics from a source protocol are stripped from routes as they are distributed into a destination protocol - therefor it is sometimes necessary to provide them manually. In the case of routes going INTO EIGRP, you need to define a 'seed metric' to give the incoming routes a baseline of metrics that EIGRP can understand. In the case of OSPF, you *can* manually define a metric if you want - but OSPF can also apply a 'default metric' to incoming, redistributed routes if none are defined manually.

"Similarly, 100 subnets is specified because that will allow all individual subnets to be sent instead of a summarized route? Is it right?"

Essentially, yes. When OSPF was first released it was a 'classful' protocol. It did not 'understand' VLSM or CIDR. The 'subnets' command was added later to allow OSPF to recognize classless subnets.

Now if your question meant to single out the number '100' as some sort of subnet count - then no. That '100' refers to the EIGRP process number, e.g.

router eigrp 100

Since it is possible to run more than one EIGRP (or OSPF, etc.) process on a device, it is necessary to distinguish which process ID you wish to distribute routes from.
Hopefully that adds a little more clarity.

[dbeattie]

You have to ask yourself does it really matter. What is the point of setting a reliability attribute to 100% or 50% if nothing ever uses this attribute to affect routing? In these cases, just set a value (normally the 100% value ) and stick to it. It really doesn't matter. By default, the only EIGRP values that matter are bandwidth and delay. This is one of the main reasons that people move away from EIGRP to OSPF, where bandwidth (OK- link cost, which is related to bandwidth) is the only relevant metric.

Hope this helps,

Dave