[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ms-ospf
Subject:    Re: network lsa in the route computation
From:       Yasuhiro Ohara <yasu () SFC ! WIDE ! AD ! JP>
Date:       1999-08-05 10:59:04
[Download RAW message or body]


thank you very much, Mr. zinin, I understand.

        yasu

From: Alex Zinin <zinin@AMT.RU>
Subject: Re: network lsa in the route computation
Date: Thu, 5 Aug 1999 14:06:34 +0400

zinin> Yasu-
zinin>
zinin> At 18:46 05.08.99 +0900, you wrote:
zinin> >Hi, Thank you for your response, Mr. Kimo
zinin> >
zinin> >I understand what you're going to say bellow. but I have never seen
zinin> >descriptions saying "link state database must not include LSA that
zinin> >will not be used in calculation" in RFC2328.
zinin>
zinin> Yes, we're allowed to have LSAs that are never referenced in
zinin> situations when originating routers are not operational or just
zinin> unreachable and cannot flush their LSAs from the domain.
zinin> This is because an LSA can be prematurely aged only by
zinin> the originating router. Also it is very possible that if an unreachable
zinin> router gets connected, the majority of LSAs are the same and
zinin> do not need to be updated.
zinin>
zinin> >
zinin> >I'm not complain, I'm just curious about this situation.
zinin> >
zinin> >Is there another reason to do premature aging? Or does it mean simply
zinin> >OSPF spec comes from policy that the memory size is more important
zinin> >than network traffic size?
zinin>
zinin> It is always a good idea to remove invalid LSAs if possible,
zinin> i.e., if the router can determine that these LSAs were originated
zinin> by itself. This is not only a question of memory, but of CPU and
zinin> BW as well, i.e., invalid LSAs are still considered when LSDB
zinin> lookup or periodical check is performed, they are still sent to
zinin> the new routers during the synchronization process, etc.
zinin>
zinin> The process of flushing invalid self-originated LSAs is
zinin> described in 13.4 and network-LSAs are just a particular case.
zinin>
zinin> Just my view.
zinin> Alex.
zinin>
zinin> >
zinin> >        yasu
zinin> >
zinin> >From: Kimo Johnson <kj@MARS.IOL.UNH.EDU>
zinin> >Subject: Re: network lsa in the route computation
zinin> >Date: Wed, 4 Aug 1999 11:31:08 -0400
zinin> >
zinin> >kj> > RFC2328 16."Calculation of the routing table" says...
zinin> >kj> >
zinin> >kj> >         The lookup process may return an LSA whose LS age is equal to
zinin> >kj> >         MaxAge. Such an LSA should not be used in the routing table
zinin> >kj> >         culculation, and is treated just as if the lookup process had
zinin> >kj> >         failed.
zinin> >kj> >
zinin> >kj> >
zinin> >kj> > jerome> when i get the network lsa with partial informations, should
zinin> i check that
zinin> >kj> > jerome> the age isnt MaxAge ?
zinin> >kj> >
zinin> >kj> > so, I think you should.
zinin> >kj> >
zinin> >kj> > I have futher question related to this. When router that had been DR
zinin> >kj> > restarts, the router will receive Network LSA that the router don't
zinin> >kj> > need this anymore. Such kind of LSA will not be used because of above
zinin> >kj> > reason, I think doing premature aging is not needed.
zinin> >kj> >
zinin> >kj> > why RFC2328 specifies to do premature aging?
zinin> >kj> >
zinin> >kj>
zinin> >kj> Yasuhiro,
zinin> >kj>         The link state database describes the collection of routers and
zinin> >kj> networks in an area, as well as networks external to the area and external
zinin> >kj> routes etc.
zinin> >kj>         The network-LSA from the router that had been DR is no longer
zinin> >kj> valid and does not describe the present topology of the network.  It
zinin> >kj> should therefore be removed from the Link State Database.  The network-LSA
zinin> >kj> from the new DR is valid because it describes the network in its present
zinin> >kj> state.
zinin> >kj>         If the router that had been DR, in your case above, does not age
zinin> >kj> its LSA prematurely, there will be 2 network-LSAs for the network until
zinin> >kj> the "bad" LSA reaches MaxAge - This could take about an hour.
zinin> >kj>
zinin> >kj> I hope this answers your question,
zinin> >kj> Kimo
zinin> >kj> +-----------------------------------------------------------+
zinin> >kj> | Kimo Johnson                      UNH InterOperability Lab|
zinin> >kj> | IP/Routing Consortium Engineer    7 Leavitt Lane, Room 106|
zinin> >kj> |                                   Durham, NH 03824-3525   |
zinin> >kj> | Tel:   (603) 862-0090                                     |
zinin> >kj> | Fax:   (603) 862-4181             Web: www.iol.unh.edu    |
zinin> >kj> | Email: kj@iol.unh.edu                                     |
zinin> >kj> +-----------------------------------------------------------+
zinin> >kj>
zinin> >
zinin> >
zinin> ------------------------------------------------------------------
zinin> Alex D. Zinin, Consultant
zinin> CCSI #98966
zinin> CCIE #4015
zinin>

[prev in list] [next in list] [prev in thread] [next in thread]