Hi Chuck,
Thanks for the OSPF tips and techniques. Your insights are very valuable as
always, and very well written, as always. I have the flu and cannot twiddle
bits in this condition, but I do have one other point to make, nonetheless. ;-)
My personal preference is to be careful with the OSPF network statements
because I get squeamish when I see all those OSPF multicast hellos going
out every 10 seconds to LANs that only have end nodes on them. So I make
the network statement very precise to avoid this, and then, if necessary,
add a command such as "redistribute connected subnets" to make sure the
rest of the world does learn about the LANs, even though they are not part
of the OSPF area.
I have never confirmed if this is a good network design practice or not and
appreciate any feedback. Thanks.
Priscilla
At 07:14 PM 7/18/00, Chuck Larrieu wrote:
>And that indeed is the trick, and the point I am trying to make.
>
>WARNING: BEGIN LECTURE - OSPF and OTHER NETWORK STATEMENTS
>
>Although it is sometimes hard to discern this from the various study
>materials ( I have not read Tom Thomas' Designing OSPF Networks, so I can't
>comment there ) the NETWORK statement in OSPF does something a bit different
>than it does in RIP, IGRP, and EIGRP. In these latter protocols, the network
>statement places the classful network into the routing process. With OSPF,
>an interface is placed into the OSPF routing process.
>
>What I believe happens is that the IOS code is written in such a way that it
>looks at the network statements, compares the contents to the contents of
>the interface address/mask table, and if there is a match, takes the
>appropriate information, and places it into one of the OSPF tables. Recall
>that OSPF is a classless protocol, supporting CIDR and VLSM, and that it
>carries subnet mask / prefix information as a part of the network
>advertisements.
>
>Therefore, the following network statements accomplish the exact same thing,
>as far as the OSPF process is concerned
>
>Network 172.16.1.1 0.0.0.0 area 0
>Network 172.16.1.1 0.0.255.255 area 0
>Network 172.16.1.1 0.0.1.1 area 0 ( why? )
>Network 0.0.1.1 255.255.0.0 area 0 ( this is the kind of thing Howard warns
>about :-> )
>Network 172.16.1.1 172.16.1.1 area 0 ( interesting pneumonic, is it not? )
>
>There are a whole lot more ways to do this, as you can see.
>
>And these all work, whether the subnet mask is /16 or /24 or /20 or /30 or
>whatever.
>
>In terms of best practice, one consideration is maintenance and ease of
>troubleshooting. Now in a core routers with 72 interfaces, al of which are
>in the OSPF process, in terms of ease of management and troubleshooting,
>which is it better to do - manually enter 72 network statements? Devise some
>clever mask or series of masks? Or just place all of them into the process
>with an all inclusive statement? Where are errors more likely to occur?
>
>The point about troubleshooting is one well worth considering. In terms of
>simplicity, the more apparent something is, the more apparent something is
>wrong, if it is wrong. But one day, sooner or later, we will all run into
>the situation where some clever person has done things cleverly, and left no
>documentation. Without a thorough understanding about how binary works, how
>the bits match up, what they do and do not match, we will be unable to
>quickly determine if things or processes or access-lists or whatever are
>functioning as they are intended.
>
>So yes, this might be considered showboating. It might be considered a silly
>trick. Or it might be considered another means of broadening one's
>understanding. It all depends upon one's outcomes.
>
>The books I've read all say to do something like network 172.16.0.0
>0.0.255.255 area 0, using the inverse network mask. This leads one to
>believe one is placing entire logical networks into the process. But if the
>mask were /21, it might be far to easy to incorrectly determine the
>appropriate mask ( 0.0.7.255 0 and thus create an evening of
>troubleshooting.
>
>END OF LECTURE
>
>-----Original Message-----
>From: Fred Ingham [mailto:[EMAIL PROTECTED]]
>Sent: Tuesday, July 04, 2000 9:09 PM
>To: Chuck Larrieu
>Subject: Re: More puzzles - OSPF Network statements
>
>Since OSPF uses the network statements to decide which
>interfaces are included in the area the wildcard mask doesn't have to
>match the interface SNM.
>
>network 0.0.0.0 255.255.255.255 area 0.0.0.0
>
>will get all interfaces in area 0.0.0.0. I assume you had
>something else in mind?
>
>Chuck Larrieu wrote:
> >
> > Gee, the recent wildcard mask puzzle was so much fun, let's do it again.
> >
> > The problem revolves around the OSPF network statement. Given an OSPF
> > routing process, what are the fewest number of network statements that
>will
> > incorporate the following networks into the given OSPF process?
> >
> > For example:
> >
> > router OSPF 100
> > Network a.b.c.d w.x.y.z area 0.0.0.0
> > Network b.c.d.e u.v.w.x area 0.0.0.0
> > etc
> >
> > Assume that all networks are in the same area ( for simplicity sake )
> > You may also assume ( for simplicity sake ) that the particular interface
>on
> > which each particular network resides is the first host address on that
> > network. For example, on network 10.10.3.0/24, the first host address
> and
> > therefore the interface address ) is 10.10.3.1
> >
> > Networks are as follows:
> >
> > 10.10.3.0 /24
> > 10.25.17.0 /16
> > 10.51.100.192 /26
> > 172.16.0.0 /16
> > 172.16.100.0 /24
> > 172.16.210.240 /28
> > 172.27.32.0 /19
> > 192.168.0.0 /24
> > 192.168.1.0 /30
> > 192.168.207.56 /29
> >
> > remember - accolades go to the one who can enter all of these networks
>into
> > a single ospf process with the fewest possible network statements.
> >
> > Have fun!
> >
> > Chuck
> >
> > Please check out my new footers for a new age
> > 1) Altruism
> > http://www.thehungersite.com/cgi-bin/WebObjects/HungerSite
> > Please help feed hungry people worldwide. A few seconds a day can make a
> > difference to many people
> >
> > 2) Shameless Hucksterism
> > http://www.certificationzone.com
> > An excellent study focal point for all levels of certification, as well as
> > the attainment of internetworking expertise. Use my name when you
>register.
> > You get good study material and I get extra time
> >
> > ___________________________________
> > UPDATED Posting Guidelines: http://www.groupstudy.com/list/guide.html
> > FAQ, list archives, and subscription info: http://www.groupstudy.com
> > Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
>
>___________________________________
>UPDATED Posting Guidelines: http://www.groupstudy.com/list/guide.html
>FAQ, list archives, and subscription info: http://www.groupstudy.com
>Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
________________________
Priscilla Oppenheimer
http://www.priscilla.com
___________________________________
UPDATED Posting Guidelines: http://www.groupstudy.com/list/guide.html
FAQ, list archives, and subscription info: http://www.groupstudy.com
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
