HQ to DR Preferred route, BGP OSPF

[wilson_1234_2]

Our HQ to DR connection has a point to point DS3 and a connection via MPLS.

HQ has the DS3 and 6 T1s multilinked into the MPLS cloud.

DR has the DS3 and a T1 into the cloud.

There is BGP configured between the two sites and each site is distributing the local OSPF routes into the BGP.

Both HQ and DR are doing mutual distribution of BGP and OSPF.

HQ has this in the BGP config:

distance bgp 150 200 201

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

I want to know why this is happeneing, I see nothing in the config that would point to costing on the interfaces.

Is the cost from the OSPF config being distributed into BGP also and due to the serial interface being 45Mps on the DS# making this the preferred route to DR?

The serial iterface is subinterfaced.

interface Serial1/0

no ip address

encapsulation frame-relay IETF

ip route-cache flow

load-interval 30

dsu bandwidth 44210

framing c-bit

cablelength 50

clock source internal

serial restart-delay 0

frame-relay lmi-type ansi

frame-relay intf-type dce

Serial1/0 is up, line protocol is up

Hardware is M1T-T3+ pa

Description: connected to MCI DS3 Disaster Recovery

MTU 4470 bytes, BW 44210 Kbit, DLY 200 usec,

reliability 255/255, txload 27/255, rxload 2/255

Does this set the Bandwiith on the interface:

dsu bandwidth 44210

[wilson_1234_2]

Thanks for the reply,

I see no filtering and ours does not seem to have any problems at the moment.

Below is out config on the edge router:

router ospf 1

router-id 10.5.70.1

log-adjacency-changes

redistribute static subnets route-map STATICtoOSPF

redistribute bgp 65001 subnets

passive-interface Loopback0

network 10.1.7.0 0.0.0.255 area 0

network 172.16.1.1 0.0.0.0 area 0

default-information originate

!

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 10.1.7.0 mask 255.255.255.0

network 10.50.0.0 mask 255.255.0.0

network 172.16.1.1 mask 255.255.255.255

redistribute static route-map STATICtoBGP

redistribute ospf 1 match internal external 2

neighbor 1.1.254.66 remote-as 65011

neighbor 1.1.254.66 version 4

neighbor 1.1.2.66 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

Can you show me an example of how you have your mutual distribution filtered?

[Edison Ortiz]

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?

Sorry, I don't understand this question.

If the HQ Edge Router is doing the resdistribution, all routers running the same protocol as the HQ Edge Router will receive those routes unless they are filtered.

When performing redistribution, you always run into the risk of creating a route-loop. It's very hard to determine if you are going to encounter a route-loop without viewing the whole network topology.

It would be ideal if you can provide a diagram depicting the whole routing topology.

__

Edison.

[wilson_1234_2]

Here is a drawing with OSPF/BGP config shown.

My question regarding the Distribution, was would I see the routes as distributed in the current route tables in the 7206 and the 3725 edge routers, or would they only show up in the adjacent routers with the edge routers only doing the distribution?

For example if the 7206 is distributing the DR BGP subnets into OSPF, would I see that in the 7206 route tables.

Looking at the 7206, it looks like I can see the OSPF database showing the BGP routes, so the answer would be the 7206 does actually have the routes in it's own database.

[Edison Ortiz]

At HQ, the 7206 will the routes coming from DR and "typical branch" as BGP.

At HQ, routers sitting behind the 7206 will see those routes as 'redistributed OSPF routes' (E1 or E2).

Same theory holds for the 3725 router sitting at the DR site.

HTH,

__

Edison.

[wilson_1234_2]

Thanks for the reply.

With what I have shown you, do you think there is a need for filtering?

For example, why doesn't the mutual ditribution cause reditribution from one protocol to the other,

then re-redistribution

then re-re-redistribution?

Is there a mechanism (lacking a better word) to prevent these endless loops of distributing?

[Edison Ortiz]

No, you don't need filtering as there isn't any concern for route-loops.

There is only one way to get into HQ and DR, that is - via the 7206 and 3725 respectively.

BGP routes are coming from the external networks and brought into OSPF for internal routing.

OSPF routes are coming from the internal networks going out as BGP routes.

This is standard practice on 95% of networks out there, that is, BGP being redistributed into an IGP.

HTH,

__

Edison.

[wilson_1234_2]

Thanks edison,

Since I have your attention a couple more questions:

What about the redistribution?

What prevents OSPF from distributing the routes into BGP and since BGP is distributing routes into OSPF, to redistribute the distributed routes again and again and again?

Am I not thinking of this correctly?

It seems something would have to prevent that from happening?

Also, what is the benifit to using BGP at the remote branches and OSPF locally?

It seems the convergence time is longer with BGP route tables.

Why not just use OSPF in the remote branches as well as in the local infrastruvture?

[Edison Ortiz]

What prevents OSPF from distributing the routes into BGP and since BGP is distributing routes into OSPF, to redistribute the distributed routes again and again and again?

Nothing prevents a routing protocol from being redistributed into another. However, per your diagram, that's not the case. You have a single point of redistribution at each location and it does not create a loop. If you were to have multiple points on redistribution on a single location, then added measures must be taken to avoid a loop. That's not the case here :)

It seems something would have to prevent that from happening?

You can implement some kind of route tagging as the routes are being brought from one protocol to another and if those routes were to encounter another point of redistribution, you can configure that router NOT to redistribute those routes based on a specific tag.

Again, not the case here based on your diagram.

Also, what is the benifit to using BGP at the remote branches and OSPF locally?

It seems the convergence time is longer with BGP route tables.

Why not just use OSPF in the remote branches as well as in the local infrastruvture?

Very good question and the answer depends on the network design and needs. BGP offers you granular routing policies and independent Autonomous System.

OSPF is an IGP and any changes in the OSPF routing table, affects everyone running the protocol.

HTH,

__

Edison.

[wilson_1234_2]

Edison, maybe I am not understanding here, and this is something I have wondered about, maybe I am not making sense.

I am wondering about the distribution in my case:

If I have BGP routes from my remote sites and they are distributed into OSPF in my edge router,

Since my edge router will have both BGP and OSPF, the BGP routes are now in OSPF from the BGP to OSPF distribution.

OSPF is distributing the same routes that were distributed from BGP back into into BGP again.

How does the router know to not go into some uncontrollable runnaway?

How does the router know to do just one distribution of BGP into OSPF and one Distribution of OSPF into BGP and stop there?

It seems that is a circle that could go out of control.

[Edison Ortiz]

The router performing the redistribution will see the BGP routes as BGP and the OSPF routes as OSPF.

Yes, the BGP routes will be in OSPF only on devices sitting behind this router. The router doing the redistribution won't change these routes on its own routing table.

For instance, perform a show ip route x.x.x.x and replace the x's for a network originated from the DR site at the 7200 router. Those routes will be shown as BGP while if you do the same in a router sitting behind the 7200 at HQ, those routes will be shown as OSPF.

Now, check for routes originate in OSPF in this same router, it will only show in the routing table as OSPF, not as BGP. Therefore, when performing a redistribution, the router check its routing table and brings the routes that matches the redistributed protocol.

HTH,

__

Edison.

[wilson_1234_2]

Edison thanks for the reply, don't give up on me yet.

This is an excellent source to learn from.

Here is an output that seems to add confusion.

My original thought during the explanations was that the distributING router did not have the distrubtED routes, which would answer the question about the runaway condition I asked about.

But earlier duing the post, I did check the DR route in OSPF database and see what I think is the route distributed from BGP on the same router, the reason fror my continued questions on this.

Perhaps you can show me what I am missing or misunderstanding.

Shown is the OSPF database of the route for the subnet residing in DR, and the BGP route as well, on the router doing the distribution, the 7206 edge to DR. The 10.5.70.1 is the router ID in OSPF for this router:

#sh ip route 10.100.4.0

Routing entry for 10.100.4.0/24

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

Tag 65011, type external

Redistributing via ospf 1

Advertised by ospf 1 subnets

Last update from 1.1.254.66 1d00h ago

Routing Descriptor Blocks:

* 1.1.254.66, from 1.1.254.66, 1d00h ago

Route metric is 2, traffic share count is 1

AS Hops 1

Route tag 65011

#sh ip ospf database external 10.100.4.0

OSPF Router with ID (10.5.70.1) (Process ID 1)

Type-5 AS External Link States

LS age: 725

Options: (No TOS-capability, DC)

LS Type: AS External Link

Link State ID: 10.100.4.0 (External Network Number )

Advertising Router: 10.5.70.1

LS Seq Number: 8000002D

Checksum: 0x7038

Length: 36

Network Mask: /24

Metric Type: 2 (Larger than any link state path)

TOS: 0

Metric: 1

Forward Address: 0.0.0.0

External Route Tag: 65011

[Edison Ortiz]

When performing redistribution, the redistributing router only looks at its routing table, NOT the particular protocol database (on this case OSPF).

The OSPF database is useful as you can verify the route is indeed being advertised out to OSPF neighbors (remember,all OSPF speaking routers running in the same area must have identical databases). However, if the route does not show as an OSPF route while issuing "show ip route x.x.x.x", then on that particular router, the route isn't treated as OSPF. As per your example, the route 10.100.4.0 is a BGP route and is being Redistributing via ospf 1.

__________

#sh ip route 10.100.4.0

Routing entry for 10.100.4.0/24

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

Tag 65011, type external

Redistributing via ospf 1

Advertised by ospf 1 subnets

Last update from 1.1.254.66 1d00h ago

Routing Descriptor Blocks:

* 1.1.254.66, from 1.1.254.66, 1d00h ago

Route metric is 2, traffic share count is 1

AS Hops 1

Route tag 65011

_________

HTH,

__

Edison.

[wilson_1234_2]

Excellent replys edison.

You have answered numerous questions that have been bugging me.

Thank you so much!

I have more questions regarding what I see in the above outputs if you are up for it.

If not, I understand.

[Edison Ortiz]

Ask away :)

[wilson_1234_2]

On the "routing descriptor blocks" below,

This line is telling me (from the star)that it is the best route to 10.100.40.0, is that correct?

In BGP, if there were numerous routes, would they show us as well with the asterisc route?

Why is the route metric 2?

#sh ip route 10.100.4.0

Routing entry for 10.100.4.0/24

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

Tag 65011, type external

Redistributing via ospf 1

Advertised by ospf 1 subnets

Last update from 1.1.254.66 1d00h ago

Routing Descriptor Blocks:

* 1.1.254.66, from 1.1.254.66, 1d00h ago

Route metric is 2, traffic share count is 1

AS Hops 1

Route tag 65011

On the below output, is the "LS Type: AS External Link" derived from the route being a distributed route?

Why is the forwarding address 0.0.0.0?

Where can I find informa on these types of things? There doesn't seem to be any resources on reading the OSPD database.

#sh ip ospf database external 10.100.4.0

OSPF Router with ID (10.5.70.1) (Process ID 1)

Type-5 AS External Link States

LS age: 725

Options: (No TOS-capability, DC)

LS Type: AS External Link

Link State ID: 10.100.4.0 (External Network Number )

Advertising Router: 10.5.70.1

LS Seq Number: 8000002D

Checksum: 0x7038

Length: 36

Network Mask: /24

Metric Type: 2 (Larger than any link state path)

TOS: 0

Metric: 1

Forward Address: 0.0.0.0

External Route Tag: 65011