From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Fri Oct 03 2003 - 21:11:28 GMT-3
At 7:30 PM -0400 10/3/03, Thomas Larus wrote:
>I think that's what it amounts to. Still, I keep waiting for someone to
>chime in with some point we have not yet considered, or some weird
>hypothetical situation.
Well, I'll chime in with something that is hypothetical in the
certification context: best practices for real production networks.
Let's say your network has one redistribution point today. Can you
guarantee that no one is ever going to install another one? I try
both to idiot-proof and genius-proof my implementations -- I want
redistribution to fail safely.
Another consideration is the potential that subsets of the address
space on one side or the other may migrate to the other side.
Especially if one or both of the IGPs is classful, this could cause
real confusion. In a world of mergers and acquisitions, this often
happens. Often causing even more confusion than acquisition is
divestiture, where a subsidiary is sold off but takes more-specifics
of your address space with it.
Mind you, in real-world designs, I go to great lengths to avoid
mutual redistribution. In most cases, I can use hierarchical
redistribution: advertise your local routes to the more-capable
protocol, and accept default from it.
In complex real-world networks, I often find it far more manageable
to use hierarchical redistribution to a "backbone of backbones", the
latter often being BGP-based. BGP gives me a wider range of policy
tools than IGPs, iBGP can handle more routes, and I have the complex
redistribution taking place in a limited set of routers. In other
words, I am much more likely to set up a "DMZ-ish" border between two
IGPs than to redistribute directly.
Obviously, there's the caveat that there will be exceptions to every rule.
>
>
>
>----- Original Message -----
>From: "Peng Zheng" <zpnist@yahoo.com>
>To: "Thomas Larus" <tlarus@cox.net>
>Cc: <ccielab@groupstudy.com>
>Sent: Friday, October 03, 2003 7:12 PM
>Subject: Re: redistribute between IGPs
>
>
>> But in that case, is it also a solution to enable ip
>> split-horizon?
>>
>> So, I will say: I don't need to do the filter with
>> the sole point of redistribution at all.
>>
>> --- Thomas Larus <tlarus@cox.net> wrote:
>> > Exactly.
>> >
>> > Tom
>> > ----- Original Message -----
>> > From: "Peng Zheng" <zpnist@yahoo.com>
>> > To: "Thomas Larus" <tlarus@cox.net>
>> > Sent: Friday, October 03, 2003 4:40 PM
>> > Subject: Re: redistribute between IGPs
>> >
>> >
>> > > Like sometimes RIP or EIGRP running on physical
>> > frame
>> > > relay interface. Right?
>> > >
>> > >
>> > > --- Thomas Larus <tlarus@cox.net> wrote:
>> > > > Okay, I am going out on a limb here, but I
>> > believe
>> > > > that, if all else in the
>> > > > whole scenario is configured just right, there
>> > > > should no need for filtering
>> > > > at the sole point of redistribution. This is
>> > > > because of the loop prevention
>> > > > behavior that has been discussed often here,
>> > that a
>> > > > router will not
>> > > > redistribute routes from one routing protocol to
>> > > > another that were just
>> > > > redistributed into the first routing protocol on
>> > the
>> > > > same router. (Often
>> > > > discussed in regard to connected routes). This
>> > is
>> > > > why you need to
>> > > > redistribute routes separately from each
>> > protocol
>> > > > into each other protocol
>> > > > on a router where you have three protocols
>> > running
>> > > > and want lots of wild and
>> > > > crazy redistribution.
>> > > >
>> > > > You sometimes see filters in single-point route
>> > > > redistribution examples
>> > > > because things are not always configured just
>> > right.
>> > > > For example, I believe
>> > > > that if you had a RIP domain in which split
>> > horizon
>> > > > was disabled somewhere,
>> > > > the split horizon problem could cause serious
>> > > > problems that might look like
>> > > > a "route feedback" problem.
>> > > >
>> > > > Tom Larus
>> > > >
>> > > > ---- Original Message -----
>> > > > From: "Peng Zheng" <zpnist@yahoo.com>
> > > > > To: <ccielab@groupstudy.com>
>> > > > Sent: Friday, October 03, 2003 3:24 PM
>> > > > Subject: redistribute between IGPs
>> > > >
>> > > >
>> > > > > Is there any scenario that routing filter is
>> > > > necessary
>> > > > > with one point redistribution between any two
>> > of
>> > > > IGPs?
>> > > > >
>> > > > > Thanks.
>> > > > >
>> > > > >
>> > > > > __________________________________
>> > > > > Do you Yahoo!?
>> > > > > The New Yahoo! Shopping - with improved
>> > product
>> > > > search
>> > > > > http://shopping.yahoo.com
>> > > > >
>> > > > > ***Get your CCIE and a FREE vacation:
>> > > > Shop.GroupStudy.com***
>> > > > >
>> > > >
>> > >
>> >
>> _______________________________________________________________________
>> > > > > Please help support GroupStudy by purchasing
>> > your
>> > > > study materials from:
>> > > > > shop.groupstudy.com
>> > > > >
>> > > > > Subscription information may be found at:
>> > > > > http://www.groupstudy.com/list/CCIELab.html
>> > > >
>> > > >
>> > >
>> > >
>> > > __________________________________
>> > > Do you Yahoo!?
>> > > The New Yahoo! Shopping - with improved product
>> > search
>> > > http://shopping.yahoo.com
>> > >
>> >
>>
>>
>> __________________________________
>> Do you Yahoo!?
>> The New Yahoo! Shopping - with improved product search
>> http://shopping.yahoo.com
>>
>> ***Get your CCIE and a FREE vacation: Shop.GroupStudy.com***
>> _______________________________________________________________________
>> Please help support GroupStudy by purchasing your study materials from:
>> shop.groupstudy.com
>>
>> Subscription information may be found at:
>> http://www.groupstudy.com/list/CCIELab.html
>
>***Get your CCIE and a FREE vacation: Shop.GroupStudy.com***
>_______________________________________________________________________
>Please help support GroupStudy by purchasing your study materials from:
>shop.groupstudy.com
>
>Subscription information may be found at:
>http://www.groupstudy.com/list/CCIELab.html
***Get your CCIE and a FREE vacation: Shop.GroupStudy.com***
This archive was generated by hypermail 2.1.4 : Mon Nov 24 2003 - 07:52:57 GMT-3