From: Curtis Phillips (cphillips@xxxxxxxxxx)
Date: Sat May 11 2002 - 21:18:59 GMT-3
Brian,
I follow what you are saying ok with regard to this.
I guess the complication comes in having a dialer map on the non-watching
side.
I maintain though based on my testing that if OSPF is permitted to run
unchecked on the BRI interfaces that
it will keep the line up.
I would add that the use of making the routing protocol traffic
uninteresting in both EIRGP and IGRP support this claim.
I looked at the following link to try to get more information.
http://www.cisco.com/warp/public/793/access_dial/bri_isdn_11049.html
It appears to be a an incomplete config though as there is an access list
101 statement that denys OSPF but doesnt seem to be applied anywhere.
My topology is identical with the exceptions of addressing and the non-use
of dialer interfaces.
I have been unsuccessful in getting the link to stay down without providing
a demand-circuit statement.
I would also add that we used a demand circuit with dialer-watch in Caslows
NMC-2 class recently.
OR as Ludwig Morales suggests you can deny OSPF in the dialer-group/list as
they appear to have started to do in the link I refer to. I am inclined to
think that will work as we would use the dialer watch to trigger and then
allow the OPSF to synchonize over the BRI even though it can not initiate
the call.
----- Original Message -----
From: "Brian McGahan" <brian@cyscoexpert.com>
To: "'Curtis Phillips'" <cphillips@suscom.net>; "'ying chang'"
<ying_c@hotmail.com>; <steven.j.nelson@bt.com>; <ccielab@groupstudy.com>
Sent: Saturday, May 11, 2002 6:33 PM
Subject: RE: Justification for use of Dialer-watch in OSPF
> Curtis,
>
> You don't need interesting traffic (dialer-list) for
> dialer-watch to work. The side that is watching will always initiate
> the call, which means that the remote side does not need interesting
> traffic defined, or for that matter even a dial string. Dialer watch,
> however, is more usefully applied to other protocols besides OSPF and
> RIP. Both OSPF an RIP have demand circuit functions, and neither IGRP,
> EIGRP, nor IS-IS have such a function. Dialer watch and demand circuit
> have obvious differences in speed of convergence too. With a demand
> circuit in OSPF, adjacency is established, databases are exchanged, and
> SPF is calculated. Only after this series of events does the demand
> circuit go down. With dialer watch, however, routing protocol traffic
> should not be defined as interesting. This means that when the line
> comes up due to a lost route, adjacency must first be established before
> any traffic forwarding can occur. This is really the only option for
> EIGRP and IS-IS, however, IGRP and RIP can use snapshot, and OSPF and
> RIP have demand circuits (triggered updates in RIP). The demand circuit
> of OSPF and RIP can be used in conjunction with dialer-watch, but it
> does not make much sense to do so. Dialer-watch is triggered by the
> loss of a specific route in your routing table. Demand circuit is
> triggered by a change in topology. Assuming that the route being
> watched is part of either your OSPF domain or RIP domain, it would
> effectively be performing the same function. Also, for dialer-watch to
> perform correctly, the DDR link must *never* go down unless the primary
> route comes back up. So to answer your question, you most likely
> wouldn't use dialer-watch and demand-circuit together, but there's
> always an exception to every rule.
>
> HTH
>
> Brian McGahan
> CCIE #8593
> 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
> Curtis Phillips
> Sent: Saturday, May 11, 2002 4:25 PM
> To: ying chang; steven.j.nelson@bt.com; ccielab@groupstudy.com
> Subject: Re: Justification for use of Dialer-watch in OSPF
>
> Ying, Steve,
>
> How about the non-dialer-watch side of the connection?
>
> Curtis
> ----- Original Message -----
> From: "ying chang" <ying_c@hotmail.com>
> To: <cphillips@suscom.net>; <steven.j.nelson@bt.com>;
> <ccielab@groupstudy.com>
> Sent: Saturday, May 11, 2002 4:21 PM
> Subject: Re: Justification for use of Dialer-watch in OSPF
>
>
> > Curtis,
> >
> > I have to agree with Steve. When I did dialer-watch a week or two ago,
> I
> > could take dialer-group out and the dialer-watch would work without
> any
> > problems. What triggers dialer-watch to dial is a watched missing
> route,
> it
> > has nothing to do with the interesting traffic specified in the
> > dialer-group. After the line come up, it will check if the primary
> line
> come
> > backup every idle-timeout seconds, the line would stay up as long as
> the
> > route is missing.
> >
> > Chang
> >
> >
> > >From: "Curtis Phillips" <cphillips@suscom.net>
> > >Reply-To: "Curtis Phillips" <cphillips@suscom.net>
> > >To: <steven.j.nelson@bt.com>, <ccielab@groupstudy.com>
> > >Subject: Re: Justification for use of Dialer-watch in OSPF
> > >Date: Sat, 11 May 2002 15:41:39 -0400
> > >
> > >Steve,
> > >
> > >I think you will find that demand-circuit is still required if
> dialer-watch
> > >is used with OSPF.
> > >Set it up and you will see that the OSPF will keep the ISDN up.
> > >
> > >Curtis
> > >----- Original Message -----
> > >From: <steven.j.nelson@bt.com>
> > >To: <cphillips@suscom.net>; <ccielab@groupstudy.com>
> > >Sent: Saturday, May 11, 2002 11:53 AM
> > >Subject: RE: Justification for use of Dialer-watch in OSPF
> > >
> > >
> > > > Curtis
> > > >
> > > > Think you may have answered your own question here except for one
> small
> > > > point.
> > > >
> > > > Dialer Watch and OSPF Demand circuit are not and were never meant
> to
> > >work
> > >in
> > > > conjunction with each other, they are two separate technologies
> that
> > >allow
> > > > demand (Dialer, X.25, atm and frame svc etc etc) circuits to keep
> quiet
> > > > unless a specific condition is met.
> > > >
> > > > In the case of ospf demand circuit the trigger is a change in
> topology
> > >etc
> > > > etc
> > > >
> > > > In the case of Dialer Watch when a watched route dissapears from
> the
> > >routing
> > > > table
> > > >
> > > > Don't confuse the two and use them as they are meant separately...
> > > >
> > > > HTH, ant more questions mail me off line...
> > > >
> > > > All the best
> > > >
> > > > Steve
> > > >
> > > > -----Original Message-----
> > > > From: Curtis Phillips [mailto:cphillips@suscom.net]
> > > > Sent: 11 May 2002 16:29
> > > > To: ccielab
> > > > Subject: Justification for use of Dialer-watch in OSPF
> > > >
> > > >
> > > > Hello All,
> > > >
> > > > I was thinking about the use of Dialer-watch with OSPF over ISDN.
> I
> know
> > > > that
> > > > demand-circuits are used in conjunction with dialer-watch in the
> same
> > >manner
> > > > as protocol packets are filtered by dialer-lists fro EIGRP or
> IGRP.
> > > >
> > > > Demand interfaces allow the dialer to bypass keep-alives and to
> dial
> > >only
> > > > when there is a change (addition or deletion) of LSA in the OSPF
> > >database.
> > > >
> > > > It seems to me that demand circuits would adequately cover the
> > >suppression
> > > > of
> > > > dial related to anything other than a change in database LSAs. So,
> I
> > >wonder
> > > > what the perceived benefit is of using dialer-watch with OSPF.
> Unless
> it
> > >is
> > > > to
> > > > track routes that are not in the OSPF process.
> > > >
> > > > Any thoughts?
> > > >
> > > > Thanks,
> > > >
> > > > Curtis
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:58:54 GMT-3