BGP/EIGRP Redistribution - BGP Backdoor

[joshehaos]

I've attached a pic of what I found off the web.  Its the same exact same setup that we have.  What I'm trying to get around is BGP routes getting preferred over internal EIGRP routes between CE1 and CE2.  We are going to be doing mutal redistribution on both CE1 and CE2 and I know the routers will prefer BGP (redistributed EIGRP routes) since that AD will be 20.  I've read on the BGP backdoor and that solution sounds great except the fact that we have 50 or some subnets at each location.  Having to add 50 network statements just doesn't seem like a good solution.  I thought about modifying the AD of internal EIGRP routes to 15.  Seeing if anyone else has done something similar to this.

Thanks

[Vasilii Mikhailovskii]

Hello.

Why don't you run BGP between CE1/CE2?

Are you doing any redistribution between protocols?

[joshehaos]

The eBGP session isn't in the same subnet as the pic shows.  They're just simply pairing with the provider at different locations.  We're doing redistribution between EIGRP -> BGP and BGP -> EIGRP.

[paul driver]

Hello

You need to modify eigrp  internal & external ad values to acommplish that

router eigrp xx

distance eigrp 9 19

res

Paul

Please don't forget to rate any posts that have been helpful.

Thanks.

[Vasilii Mikhailovskii]

Are you using CE1-CE2 link as WAN link, or it's LAN?

---

Distance 9/19 could cause routing loop (if filtering is not in place):

CE2 announce to PE2, PE2 -> PE1,

PE1 has BGP entries from CE1 and PE2.

If route flaps, EIGRP convergence is faster than BGP.

Situation could happen that PE1 has entry from PE2 only and starts announcing to CE1.

CE1 announces to CE2, CE2 announces to PE2, PE2 announces to PE1.

So, I would say:  redistribute into BGP only internal EIGRP and adjust AD to 9/170.

---

There also could be an issue that ISP will choose CE2 as a best path for subnets that are located over CE1 (and vice versa).

[joshehaos]

Thanks for the input guys.