From: Yigit Zorlu (alec_cisco@yahoo.com)
Date: Mon Oct 09 2000 - 05:51:37 GMT-3
Ciao,
AFAIK, since area 0 is not active in any of the interfaces (i.e. no networks
defined in area 0) sh ip ospf data displays that area 0 is down and that may
cause problems.
What about creating two different processes on R12 one should have area 1
other can have area 2 networks since we have no networks in area 0 ? Then we
can redistribute to eachother...
Marco
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
Howard C. Berkowitz
Sent: Thursday, October 03, 2002 3:06 PM
To: ccielab@groupstudy.com
Subject: Re: OSPF Design
At 7:58 AM -0300 10/3/02, Carlos G Mendioroz wrote:
>All,
>what's wrong with two VLs ?
>This would actually make R12 an ABR (it is not unless
>it has a link in the backbone). Then it will do (if it
>is the best path, which I would guess is the case)
>route between areas 1 & 2.
In general, I don't like to use virtual links unless I have to. This
was long a recommendation in CID. There are both practical and
protocol considerations.
Practically, there were bugs in the OSPFv1 specification of VLs, so
people stayed away from them, and the code just wasn't tested to the
same extent as other features. The subtlety of configuring a router
ID rather than an interface as an endpoint can make them hard to
troubleshoot, as well as authentication.
I'm also assuming that the best path is between the two non-backbone
areas. The static route with OSPF backup approach is something I've
primarily used to force traffic onto a preferred path -- manual
traffic engineering.
From the protocol standpoint, running virtual links in an area makes
it ineligible for any kind of stubbiness.
>
>Jason Sinclair wrote:
>>
>> Howard,
>>
>> I agree - makes much more sense design wise and will actually work as
>> expected.
>>
> > Regards,
>\
> >
>> -----Original Message-----
>> From: Howard C. Berkowitz [mailto:hcb@gettcomm.com]
>> Sent: Thursday, 3 October 2002 10:10
>> To: ccielab@groupstudy.com
>> Subject: RE: OSPF Design
>>
>> At 9:49 AM +1000 10/3/02, Jason Sinclair wrote:
>> >I would not call this design as such, however this could be configured
via
>> >the use of two virtual links to area 0 via the other two respective
areas.
>>
>> Mea culpa, Jason. Let me suggest a compromise that does force an ABR
>> area 0.0.0.0 connection, but doesn't send this traffic over it unless
>> the primary link fails.
>>
>> What I usually do in a case like this is define a static route(s)
>> between the endpoints, give it an administrative distance lower than
>> OSPF's, and ABSOLUTELY NOT REDISTRIBUTE IT INTO OSPF. That way,
>> traffic will take the preferred path as long as it's up, but will use
>> area 0.0.0.0 as backup if there is a primary path failure.
>>
>> >au
>> >
>> > -----Original Message-----
>> >From: Howard C. Berkowitz [mailto:hcb@gettcomm.com]
>> >Sent: Thursday, 3 October 2002 04:28
>> >To: ccielab@groupstudy.com
>> >Subject: Re: OSPF Design
>> >
>> >At 1:51 PM -0400 10/2/02, Peter Wodle wrote:
>> >>I wonder if the following OSPF Design is valid/workable?
>> >>
>> >>Have the area 0 with a no of routers. 1 ABR links area 0 to Area 1.
>> >>another ABR links area 0 to Area 2. So far no issues. But then we
>> >>have R12 that connects area 1 & area 2 togather i.e. R12 has one
>> >>interface in area 1 and one one interface in area 2. R12 has no
>> >>interface in area 0. So, it is an ABR (right?) but has no link to
>> >>area 0. Could this could lead to area 1 & area 2 exhange routes via
>> >>this "backdoor" router?R12
>> >>
>> >>Can we do this? I can see it is bad design but...
>> >
>> >Not in the present standard or most (if not all) Cisco
>> >implementations. There is a proposal in the IETF OSPF Working Group
>> >to do something close to it:
> > >http://www.ietf.org/internet-drafts/draft-ietf-ospf-mlinks-03.txt
This archive was generated by hypermail 2.1.4 : Tue Nov 05 2002 - 08:35:41 GMT-3