From: Brian McGahan (bmcgahan@internetworkexpert.com)
Date: Fri Nov 21 2003 - 15:59:03 GMT-3
Cisco's OSPF implementation (not sure about the standard) has a
special network type to deal with this problem. This is this network
type 'point-to-multipoint non-broadcast' Typically it is used in a NBMA
scenario when you have one logical network that has VCs of varying
speeds, it can be applied to this problem as well.
Suppose we have the following:
_X_
R1 R2
_|____|_
|
R3
R1, R2, and R3 share a broadcast Ethernet segment. R1 and R3
connect to the segment with FastEthernet interface, while R2 uses a
regular 10Mbps Ethernet interface. R1 and R2 are attached to
destination "X". From the perspective of R3, the cost to destination X
is equal through R1 and R2 (cost of 1 for FastE on R3). However, since
R2 only has 10Mb in actuality it's not. Changing the cost of R3's
interface isn't going to affect anything because as you mentioned
before, it connects to the same segment. With network type P2M
non-broadcast, you can specify the cost on a per neighbor basis:
R3:
interface Ethernet0/0
ip ospf network point-to-multipoint non-broadcast
!
router ospf 1
neighbor 10.0.0.1 cost 1
neighbor 10.0.0.2 cost 10
This way R3 knows that prefixes learned from R2 should be offset
with a cost of 10 instead of the interface cost of 1.
HTH,
Brian McGahan, CCIE #8593
bmcgahan@internetworkexpert.com
Internetwork Expert, Inc.
http://www.InternetworkExpert.com
Toll Free: 877-224-8987
Direct: 708-362-1418 (Outside the US and Canada)
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf
Of
> Howard C. Berkowitz
> Sent: Friday, November 21, 2003 10:33 AM
> To: ccielab@groupstudy.com
> Subject: Re: tune ospf cost in a broadcast media
>
> At 7:41 AM -0800 11/21/03, zzk wrote:
> >Hi
> >I understand in a point to point link, it is better to
> >change 'ip ospf cost' to the same value on both sides.
> >However if in a broadcast media (e.g. multiple routers
> >in a backbone vlan), if we change 'ip ospf cost' on
> >one router interface only, what will be the impact?
> >I feel ospf is a link state protocol, and all routers
>
> Setting different costs, as opposed to different timer values, won't
> break anything. I agree uniformity is better for troubleshooting, but
> there very well may be reasons to have different costs.
>
> Now, for your more specific example, with a different interface cost
> on one router connected to a multiaccess link, does the router have
> additional links that it can use to reach the same destination? In
> that case, having a higher cost on the less-preferred link may be
> perfectly valid means of expressing an intra-area routing policy.
>
>
This archive was generated by hypermail 2.1.4 : Fri Dec 12 2003 - 12:29:15 GMT-3