Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. And see here for current known issues.

New Member

OSPF auto-cost reference-bandwidth 100000

Is this a 100 gig reference cost?

Are there any drawbacks to going this high?

2 ACCEPTED SOLUTIONS

Accepted Solutions
Hall of Fame Super Silver

Re: OSPF auto-cost reference-bandwidth 100000

Hello Andrew,

yes in this way an ip ospf cost 1 is associated to a 100GE link.

the link OSPF cost is an unsigned 16 bit quantity.

in this way a link with maximum cost 65535 is a link with bandwidth 1,5 Mbps.

So you haven't in your network links with speeds less then T1 you should be fine.

If there are their cost is the maximum possible or the lowest 16 bits of the ratio ?

I don't know.

If it is the lowest 16 bits it can be a problem that you can fix by setting manually the cost to 65534.

you can check with

sh ip ospf interface

to perform the test whatever interface is you can set an artifical low bandwidth then configure ospf with the autoreference-bw 1000000 and see what

sh ip ospf interface

tells about the cost.

Hope to help

Giuseppe

Super Bronze

Re: OSPF auto-cost reference-bandwidth 100000

As Giuseppe notes, the OSPF cost metric is a 16 bit value, and what can happen is slower links can overflow the metric cost. Remember, it's not just one link's metric, because the metric is cumulative from router to destination (depending on route types and area configuration).

From my experience, when the metic overflows it stays at the maximum value. What happens, if you have multiple paths to a destination, OSPF might not be able to distinguish a lower cumulative bandwidth path from a better path, i.e. they will look equal. (BTW: this is sort of the converse if you leave the default Cisco OSPF cost based on 100 Mbps, can't distiguish from 100 Mbps, from anything with higher bandwidth, e.g. multiple 100 Mbps Etherchannel, gig, etc.)

4 REPLIES
Hall of Fame Super Silver

Re: OSPF auto-cost reference-bandwidth 100000

Hello Andrew,

yes in this way an ip ospf cost 1 is associated to a 100GE link.

the link OSPF cost is an unsigned 16 bit quantity.

in this way a link with maximum cost 65535 is a link with bandwidth 1,5 Mbps.

So you haven't in your network links with speeds less then T1 you should be fine.

If there are their cost is the maximum possible or the lowest 16 bits of the ratio ?

I don't know.

If it is the lowest 16 bits it can be a problem that you can fix by setting manually the cost to 65534.

you can check with

sh ip ospf interface

to perform the test whatever interface is you can set an artifical low bandwidth then configure ospf with the autoreference-bw 1000000 and see what

sh ip ospf interface

tells about the cost.

Hope to help

Giuseppe

Super Bronze

Re: OSPF auto-cost reference-bandwidth 100000

As Giuseppe notes, the OSPF cost metric is a 16 bit value, and what can happen is slower links can overflow the metric cost. Remember, it's not just one link's metric, because the metric is cumulative from router to destination (depending on route types and area configuration).

From my experience, when the metic overflows it stays at the maximum value. What happens, if you have multiple paths to a destination, OSPF might not be able to distinguish a lower cumulative bandwidth path from a better path, i.e. they will look equal. (BTW: this is sort of the converse if you leave the default Cisco OSPF cost based on 100 Mbps, can't distiguish from 100 Mbps, from anything with higher bandwidth, e.g. multiple 100 Mbps Etherchannel, gig, etc.)

New Member

Re: OSPF auto-cost reference-bandwidth 100000

Joseph, thanks for the lesson. This helps me understand the metrics in the routing table better. Andrew

New Member

Re: OSPF auto-cost reference-bandwidth 100000

Thanks Giuseppe. That clarifies what I'm seeing in the route table. I really appreciate it.

Andrew

2885
Views
0
Helpful
4
Replies