sorry

"The HQ site preferrs the MPLS cloud for all remote branches, but preferrs the DS3 for the HQ subnets. "

This should be:

The HQ site preferrs the MPLS cloud for all remote branches, but preferrs the DS3 for the DR subnets.

Here is the BGP config for the edge router to the DR site.

There is a multilink to the MPLS cloud that has a link to DR and the serial interface for the DS3.

The 192.168.2.66 is the serial interface for the DS3.

The 1.2.1.29 is the multilink to the MPLS cloud, that goes to DR.

All subnets to DR have the next hop as 192.168.2.65, which is the DS3 interface on the DR router.

See anything that makes the DS3 preferred?

router bgp 65001

no synchronization

bgp log-neighbor-changes

network 0.0.0.0

network 10.1.0.0 mask 255.255.0.0

network 172.16.254.1 mask 255.255.255.255

redistribute static route-map STATICtoBGP

redistribute ospf 1 match internal external 2

neighbor 192.168.2.66 remote-as 65011

neighbor 192.168.2.66 version 4

neighbor 192.168.2.66 soft-reconfiguration inbound

neighbor 1.2.1.29 remote-as 65000

neighbor 1.2.1.29 version 4

neighbor 1.2.1.29 soft-reconfiguration inbound

distance bgp 150 200 201

no auto-summary

Hi there,

I think that HQ prefers the MPLS cloud for all remote branches is because is has the smaller AS path attribute.

Secondly, when redistributing OSPF into BGP, the cost is transferred into the Metric attribute of BGP (or MED - Multiple Exit Discriminator). To see that I think the command is "show ip bgp 10.x.x.0" from memory.

The distance bgp 150 200 201 is changing the administrative distance of external bgp, internal bgp and local. Which means that when there is a choice between OSPF (admin distance of 110) and external BGP (usually admin distance of 20, but now 150) the router would choose OSPF.

I hope I've helped and answered your questions to your satisfaction champ!

Brad

As Edison points out, Administrative distance does not play a part with BGP path selection

Have a look at this discussion.

http://forums.cisco.com/eforum/servlet/NetProf?page=netprof&forum=Network%20Infrastructure&topic=Routing%20and%20Switching&CommCmd=MB%3Fcmd%3Ddisplay_location%26location%3D.1dda2f44

HTH

Narayan

I agree, but what gets installed into the routing table does..

I have read all of the replys and the other thread and I still do not see it.

Here is the output of a couple of the related "sh" commands to one of the DR subnets.

This shows weight the same on both paths, local preference is the same on both paths, the only thing I see that is different is metric (not supposed to matter) and the preferred path is 2, the other path shows nothing.

The only thing I see in HQ and DR BGP config is that the remote-as is higher for the DS3 path than the MPLS path.

The MPLS cloud remote-as is 65000 for both DR and HQ.

The HQ router, the remote-as for DR via DS3 is 65011

The DR router, the remote-as for HQ vis DS3 is 65001.

Here is some information from the GQ edge router for a DR subnet:

#sh ip bgp

BGP table version is 4279, local router ID is 172.16.2.1

Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,

r RIB-failure, S Stale

Origin codes: i - IGP, e - EGP, ? - incomplete

Network Next Hop Metric LocPrf Weight Path

* 10.100.10.0/24 1.2.1.29 0 65000 65011 ?

*> 192.168.2.66 2 0 65011 ?

#sh ip bgp 10.100.40.0/24

BGP routing table entry for 10.100.40.0/24, version 89

Paths: (2 available, best #2, table Default-IP-Routing-Table)

Advertised to update-groups:

1

65000 65011, (received & used)

1.2.1.29 from 1.2.1.29 (1.24.1.128)

Origin incomplete, localpref 100, valid, external

65011, (received & used)

192.168.2.66 from 192.168.2.66 (172.16.2.5)

Origin incomplete, metric 2, localpref 100, valid, external, best

#sh ip route 10.100.40.0

Routing entry for 10.100.40.0/24

Known via "bgp 65001", distance 150, metric 2

Tag 65011, type external

Redistributing via ospf 1

Advertised by ospf 1 subnets

Last update from 192.168.2.66 7w0d ago

Routing Descriptor Blocks:

* 192.168.254.66, from 192.168.2.66, 7w0d ago

Route metric is 2, traffic share count is 1

AS Hops 1

Route tag 65011

Does anyone see anything the would answer this, or is it obvious and I missed it?

Based on your output you have 2 BGP routes in the table for network 10.100.40.0/24.

Looking at the output closer, it 'almost' look identical but let me help you spot the difference

65000 65011, (received & used)

1.2.1.29 from 1.2.1.29 (1.24.1.128)

Origin incomplete, localpref 100, valid, external

65011, (received & used)

192.168.2.66 from 192.168.2.66 (172.16.2.5)

Origin incomplete, metric 2, localpref 100, valid, external, best

Notice how the last entry only has one AS in the path vs the first entry with 2 AS's in the path?

The link I referred in the first post explains the BGP Best Path selection. Take a moment to read it, please :)

Edison,

I really appreciate your reply and knowlege.

I laughed when I saw your smiley face.

I will read it again and see if it sinks in, but sometiems it is like getting through lead.

Thanks for pointing out the difference.

Edison,

Wouldn't the shortest AS-Path be like hop count?

BGP is counting hops?

Not really. If you have an iBGP peer the AS_PATH remains the same but you are one hop away :)

You can consider the AS_PATH a 'hop count' only between eBGP routers.

__

Edison.

On my example is the "external" shown here designating the "e" in the eBGP router that you are talking about?

If so, why are they considered external?

65000 65011, (received & used)

1.2.1.29 from 1.2.1.29 (1.24.1.128)

Origin incomplete, localpref 100, valid, external

65011, (received & used)

192.168.2.66 from 192.168.2.66 (172.16.2.5)

Origin incomplete, metric 2, localpref 100, valid, external, best

eBGP peering is considered any neighbor that does not belong to the same AS.

For instance, if you have a router with BGP AS 65000 and peer with another router with BGP 65000, that's considered an iBGP peering.

While if you have a router with BGP 65001 peering with another router with BGP 65000, that's considered an eBGP peering.

Edison,

When having both OSPF and BGP doing mutual distribution from each other,

BGP would show all routes in the OSPF route table as being BGP routes, and conversly, all BGP routes would be in the OSPF route table.

If My HQ edge router is the one doing the distribution, will the distributed routes only show up in the routers adjacent to the edge router, or should they show up in the Edge router as well?

And what prevents the router (doing the mutual distribution) from continually distributing routes into each other and going into some type of runaway conditon?

Best thing is to avoid Mutual redistribution altogether. Distribute ospf into BGP and use network statement under BGP (we do something similar in our network)

If mutual redistribution is a must, you need to setup filters in such a way that a ospf learnt route is not injected back into OSPF via BGP and vice versa

HTH

Narayan