From: Howard C. Berkowitz (hcb@xxxxxxxxxxxx)
Date: Sun Jul 21 2002 - 16:36:21 GMT-3
At 12:13 PM -0400 7/21/02, Peter van Oene wrote:
>Hi Tom,
>
>Cost and metric tend to be used somewhat interchangeably in my
>experience. A slight difference might be the metrics in OSPF often
>refer to both internal and external destinations whereas costs
>generally refer to internal destinations. That is simply my
>interpretation of the rfc however. In practise, LSA's populate a
>metric field while OSPF interfaces have costs associated with them.
>
>Likely doesn't clear anything up, but that's about the best I can do :)
>
>Pete
I would characterize it that metric values are associated with hops,
while a route has a cost value. The actual units and computation of
a metric value and a route value are protocol specific.
It was a specific intention of the OSPF designers NOT to have a
standard metric computation, although 10**8/BW does appear as a
default in the OSPF MIB RFC. Not having a specific metric was
intended to give flexibility.
For example, Wellfleet assumed that anyone installing OSPF in a new
network would be converting from RIP. Therefore, their default
interface metric was 1, which meant that during the conversion, OSPF
would compute cost by hop count, simplifying troubleshooting.
Wellfleet documentation recommended manually configuring 10**8/BW
once the conversion was stable.
Another factor is that the 10**8/BW rule assumed the maximum
bandwidth would be 100 Mbps, which obviously is no longer the case.
Protocol extensions for scaling exist, but are not widespread.
>
>At 08:52 AM 7/18/2002 +0900, Tom Young wrote:
>>Shuyi:
>>
>> Your description about the bandwidth and cost was
>>right. But I want to know the metric, and the relevance
>>about metric and cost.
>>
>>Thanks
>>
>>
>>Tom
>>
>> --- Brian McGahan <brian@cyscoexpert.com> $B$+$i$N%a%C%;(B
>>$B!<%8!'(B
>>> Shuyi,
>>>
>>> The "bandwidth" value of a link doesn't have
>>> anything to do with
>>> the physical bandwidth of the interface. For
>>> example, a serial
>>> interface on a 2500 series router has a bandwidth
>>> *value* of 1544 Kbps.
>>> This does not reflect that these interfaces can
>>> physically support up to
>>> 2Mbps or 4Mbps, and whether or not you have a
>>> 256Kbps frame-relay
>>> circuit attached to it. The "bandwidth" value of an
>>> interface is
>>> static, and is set using the interface "bandwidth"
>>> command.
>>>
>>> OSPF computes it's cost value by taking
>>> (ReferenceBandwidth /
>>> InterfaceBandwith) ReferenceBandwidth has a default
>>> value of 10^8 bps,
>>> or 100Mbps. Therefore a 100Mbps FastEthernet
>>> interface has a cost of 1.
>>> The reference bandwidth value can be changed with
>>> the OSPF command "
>>> auto-cost reference-bandwidth X" where X is the
>>> reference bandwidth in
>>> terms of Mbits per second. If you want a Gigabit
>>> link to have a cost of
>>> 1, the reference bandwidth should be 1000.
>>>
>>> HTH
>>>
>>> Brian McGahan, CCIE #8593
>>> Director of Design and Implementation
>>> brian@cyscoexpert.com
>>>
>>> CyscoExpert Corporation
>>> Internetwork Consulting & Training
>>> http://www.cyscoexpert.com
>>> Voice: 847.674.3392
>>> Fax: 847.674.2625
>>>
>>>
>>> -----Original Message-----
>>> From: nobody@groupstudy.com
>>> [mailto:nobody@groupstudy.com] On Behalf Of
>>> Shuyi Li
>>> Sent: Wednesday, July 17, 2002 1:20 PM
>>> To: Jason Sinclair
>>> Cc: 'Tom Young'; ccielab@groupstudy.com
>>> Subject: RE: OSPF 's cost and metric
>>>
>>> Jason,
>>>
>>> Regarding the OSPF metric, I have a question for you
>>> that, the cost of
>>> links is considered to be the current available BW,
>>> or just the total BW
>>>
>>> even it's being occupied, say 50%. Please advise.
>>>
>>> thanks in advance.
>>> /shuyi
>>>
>>>
>>> At 05:22 PM 7/17/2002 +1000, Jason Sinclair wrote:
>>> >Tom,
>>> >
>>> >In OSPF the metric is the cost. What I mean here is
>>> best clarified as
>>> >follows:
>>> >
>>> >1. In RIP the metric as we know is hop count.
>>> >2. EIGRP/IGRP use a composite metric based on
>>> things such as bandwidth,
>>> >delay, etc
>>> >3. In OSPF the metric is based on the cost of
>>> links. The lower the cost
>>> the
>>> >more preferred the path
>>> >
>>> >Hope this makes sense.
>> > >
>>> >Regards,
>>> >
>>> >Jason Sinclair CCIE #9100
>>> >Manager, Network Control Centre
>>> >POWERTEL
>>> >55 Clarence Street,
>>> >SYDNEY NSW 2000
>>> >AUSTRALIA
>>> >office: + 61 2 8264 3820
>>> >mobile: + 61 416 105 858
>>> >email: sinclairj@powertel.com.au
>>> >
>>> >-----Original Message-----
>>> >From: Tom Young [mailto:gitsyoung@yahoo.co.jp]
>>> >Sent: Wednesday, 17 July 2002 16:35
>>> >To: ccielab@groupstudy.com
>>> >Subject: OSPF 's cost and metric
>>> >
>>> >Hi, group.
>>> >
>>> > The OSPF's "cost" and "metric" parameters made
>>> me
>>> >confused. Who can clear it for me?
>>> >
>>> >
This archive was generated by hypermail 2.1.4 : Sat Sep 07 2002 - 19:36:38 GMT-3