From: Baety Wayne SrA 18 CS/SCBX (Wayne.Baety@kadena.af.mil)
Date: Tue Oct 08 2002 - 00:04:03 GMT-3
You don't need secondaries with a FLSM routing protocol as long as the /25
that is generated and advertised in your FLSM domain doesn't belong to a /24
supernet advertised in your VLSM domain. If it does then you need to make
sure you're advertising /25s to your FLSM domain and not a /24 (for any
supernet /24s that have matching /25s in your FLSM domain). Just include
the 'subnets' keyword when redistributing your OSPF routes (assuming you
just have another /25 floating around your OSPF domain). If it really is a
bonafied /24 supernet that is being advertised in your OSPF domain then
place a secondary network on an interface pointing to a /25 in your OSPF
domain (as detailed in the previous description) on a router that is not
actually generating the /24 supernet. Make sure the interface that you
choose has a router that can proxy arp for the /25. Also, include the
'subnets' keyword when redistributing your VLSM routing protocol into a
FLSM. This way your FLSM domain will know to pass traffic that belongs to
the other /25 that it isn't generating.
The second part of your concern dealt with the issue of split horizon. I
will summarize how this could possibly relate to our issue and the reasons
why it is a non issue. There are only a couple of cases in which the
disabling of split horizon is warranted.
The first situation concerns itself with multipoint interfaces in a hub and
spoke topology. In this case, there isn't all to all connectivity with
every router on the multipoint interface. Therefore, a hub router would be
forced to advertise a network out the same interface that an advertisement
was received on in order for other spoke routers to hear about an
advertisement. Since only the hub router has full connectivity to other
spokes it is the responsibility of the hub router to disable split horizon
on that interface and advertise networks received from one spoke to all
other spokes. All the spokes would then be required to set up proper
filtering to filter out received advertisements about networks about which
they have themselves advertised. This is because in actuality, the hub
router is advertising all networks it receives from each spoke to all other
spoke routers on the the same multipoint interface (including the spoke that
originally advertised the network). Unless both your OSPF router and RIP
router are on the same multipoint interface (or sub-interface) this isn't an
issue with you.
The second situation concerns itself with interfaces in which a primary and
a secondary network are defined. In order for a router that is joined to
both the secondary network and a primary network be able to advertise the
primary network (and other networks received from the primary network) to
other routers that are part of the secondary network, split horizon needs to
be disabled on that router adjoined to both networks. Otherwise routers
joined to the secondary network will not be aware of the primary network or
any networks received from any router on the primary network. Proper
filtering would need to be applied to all other routers appearing on the
interface (both adjoined to the primary and secondary networks) so that
information that they advertise do not get accepted from the router in which
split horizon is disabled. However, we are specifically not advertising
information about the secondaries that we define in this technique to your
OSPF domain. These are phantom networks designed to advertise reachability
information into your RIP domain. Split horizon does not need to be
disabled on the router that defines these phantom networks because we will
be advertising information about the secondaries through another interface.
Once again, unless both your OSPF router and RIP router appear on the same
interface (yet differnet networks) this situation does not apply to your
problem.
Needless to say, relying on phantom networks and proxy arp isn't good design
practice.
WAYNE A. BAETY, SRA, MCSE, USAF
18th Communications Squadron
DSN: 634-9927
Commercial: 011 81 611 734 9927
-----Original Message-----
From: Rick [mailto:ccie_2003@hotmail.com]
Sent: Monday, October 07, 2002 3:22 AM
To: ccielab@groupstudy.com
Subject: FLSM and VLSM (I'm still having issues)
The secondary addressing is not working for me with this. I have a /25
connecting from my rip domain to OSPF ASBR and the rest of all routes are
/24s. Do you have turn of split horizon on both sides of the serial link and
put the secondary address on both sides? I have read all I can find about
this but I must be missing something because it is not working correctly for
me. Can some please explain the viable tools to perform this task between
VLSM and FLSM?
Thanks,
Rick
----- Original Message -----
From: "Baety Wayne SrA 18 CS/SCBX" <Wayne.Baety@kadena.af.mil>
To: <ccielab@groupstudy.com>
Sent: Thursday, October 03, 2002 4:03 AM
Subject: RE: What if OSPF routes have shorter masks than RIP?
> Answers inline (bracketed)....
>
> WAYNE A. BAETY, SRA, MCSE, USAF
> 18th Communications Squadron
> DSN: 634-9927
> Commercial: 011 81 611 734 9927
>
>
>
> -----Original Message-----
> From: Hunt Lee [mailto:huntl@webcentral.com.au]
> Sent: Thursday, October 03, 2002 3:51 PM
> To: 'Bruce Williams'
> Cc: 'Young K. Bae'; 'Chris Hugo'; 'Donny MATEO'; 'Baety Wayne SrA 18
> CS/SCBX'; 'folivore'
> Subject: RE: What if OSPF routes have shorter masks than RIP?
>
>
> Hello,
>
> I have read through the entire thread for 3 times already. However,
> I'm still not so clear on what are the 4 solutions that Wayne
> proposed, can someone please kindly explain to me??
>
> 1st Solution:
>
> On both the interfaces connecting between R1 & R2, use a secondary IP
> from the 172.16.0.x/25 network, so the RIP network will have the same
> subnet
mask
> as OSPF network (this I understand)
>
> 2nd Solution:
>
> [[[[[[[[[[
>
> Read solution 3 carefully and understand the principles before reading
this.
>
>
> RIP will automatically translate all subnet advertisements received
> from
one
> classful network into a single advertisement containing only that
> classful network when advertised to another classful network. It is
> sorta like aggregating an address with BGP, except you have no control
> over the aggregation.
>
> For instance, when 172.16.200.0/25 is advertised from one router unto
> a
RIP
> enabled interface of another router that has an address like
172.17.201.1/25
> since 172.16 is a different classful network thatn 172.17, all subnets
> pertaining to 172.16/16 (including 172.16.200.0/25) will be pruned
> from
the
> update and replaced with a single advertisement of 172.16.0.0/16.
> This is called a subnet default route. You can use this feature of
> RIP to your advantage by using a different classful network prefix in
> your RIP domain than your OSPF domain. Your OSPF domain will send
> this subnet default
route
> to your RIP domain, and your RIP domain will maintain reachability to
> that classful network.
>
> -Wayne
>
> ]]]]]]]]]]
>
> Assign it with a completely different classful (or classless??)
> network
for
> the RIP domain
>
> e.g. 192.168.1.0 /24
>
> Problem 1) I have tried to set the RIP network as a Classful network
> (192.168.1.0/24) & classless network (192.168.1.0 /28) & they both
> work... so isn't sure what Wayne means by using a "classful network
> altogether"
>
> Problem 2) I tried to turn on / off RIP's auto-summarization by "no
> auto-summary" / "auto-summry" under RIP process and once again, they
> both work too... so again wasn't sure what Wayne means by "relying on
> auto-summarization & advertisement of the classful network"
>
>
> 3rd Solution:
>
> [[[[[[[[[[[
>
> Your ospf domain had the network 172.16.200.0/25 advertised, yet your
> RIP domain only understands /26 sized subnets of 172.16/16, other
> classful networks (e.g. 172.17/16, 172.18/16, etc), and the default
> route
(0.0.0.0).
> Translate 172.16.200.0/25 into 2 different sized /26 subnets and your
> RIP domain will then understand it.
>
> This could be accomplished with this....
> (Example 1)
>
> ip route 172.16.200.0 255.255.255.192 eth0
> ip route 172.16.200.64 255.255.255.192 eth0
>
> Or This...
> (Example 2)
>
> ip route 172.16.200.0 255.255.255.192 [next-hop ip address] ip route
> 172.16.200.64 255.255.255.192 [next-hop ip address]
>
> However, as you have stated you cannot use static routes. Therefore
> slap those networks on an interface in your OSPF domain as
> secondaries. It
will
> have the same effect as making a static to an interface as shown in
Example
> #1. This will make those nextworks appear to your routing process as
being
> directly connected. That means your router will ARP for those
> networks as though they appeared on the connected medium. However,
> you've already defined that network somewhere else in your OSPF domain
> as
172.16.200.0/25.
> What will happen is that the router where on these statics are applied
> (example 1 style only) will send an ARP broadcast over the medium for
> any packets it receives falling within that range. Another router
> that sees these ARP broadcasts and has proxy-arp turned on will
> respond to these ARP broadcasts with its own interface address and
> offer to forward those
packets
> to their correct destination. This is called proxy ARPing and that's
> why you would need proxy-arp turned on.
>
> -Wayne
>
> ]]]]]]]]]]
>
> "He could also subdivide the smaller size prefix with secondary
> networks into multiple networks that match the larger subnet size of
> the RIP domain (3rd Solution). To get this into the routing table
> static addressing
would
> be the the preferred method but using secondaries on one of the router
links
> in the OSPF domain and relying on proxy-arp to determine destination
> next-hop would also work."
>
> I'm completely loss =( not sure wot Wayne is referring to here?!?
> What are the smaller size prefix / larger subnet size of RIP domain???
> And
what
> static addressing is Wayne referring to?
>
>
> 4th Solution:
>
> [[[[[[[[[[
>
> Solution 4 is exactly like solution 1 except you use a tunnel
> connection
to
> interconnect your routers thereby avoiding the necessity to "pathing"
> your routing information throughout your network.
>
> Once you research and understand how to work with just creating simple
> tunnel interfaces, you'll know what we mean. You'll get an a-ha.
>
> ]]]]]]]]]]
>
> "A 4th Solution (that was mentioned) is to create a tunneled interface
> matching the subnet size of the RIP domain with the endpoint of the
> tunnel running to each RIP router requiring reachability information
> about that subnet size."
>
> I haven't used any Tunnel interface before. Can someone kindly give
> me
some
> e.g. here? How would I do it for Young's e.g.?
>
>
> Thanks.
>
> Best Regards,
> H.
>
>
>
>
> -----Original Message-----
> From: Bruce Williams [mailto:bwilliams175@comcast.net]
> Sent: Wednesday, 2 October 2002 1:31 PM
> To: Donny MATEO
> Cc: ccielab@groupstudy.com
> Subject: RE: What if OSPF routes have shorter masks than RIP?
>
>
> Split Horizon should be disabled on the interface that has the
> secondary address so that both networks can be advertised out the
> interface.
>
> Check out this link on secondary addresses and split horizon.
> http://www.cisco.com/warp/public/105/41.html
>
> Bruce
>
> -----Original Message-----
> From: Donny MATEO [mailto:donny.mateo@sg.ca-indosuez.com]
> Sent: Tuesday, October 01, 2002 11:12 PM
> To: Bruce Williams
> Cc: ccielab@groupstudy.com
> Subject: RE: What if OSPF routes have shorter masks than RIP?
>
>
>
> Could you be more specific on which issue. If it's a routing feedback
> from the other router, for most interface split-horizon is enabled.
> The only execption i know and had accountered so far is the physical
> serial
interface
> for frame-relay. Do let me know if you found other issues. I'll try to
> lab it again tonight.
>
> Donny
>
>
>
> Bruce Williams
> <bwilliams175@com To: Donny MATEO
> <donny.mateo@sg.ca-indosuez.com>, Richard Davidson
> cast.net> <rich@myhomemail.net>
> Sent by: cc:
> ccielab@groupstudy.com
> nobody@groupstudy Subject: RE: What if OSPF
> routes have shorter masks than RIP?
> .com
>
>
> 02-10-2002 10:37
> Please respond to
> Bruce Williams
>
>
>
>
>
>
> If you use secondary addresses, how do you deal with the split horizon
> issue?
>
> Bruce
>
>
> -----Original Message-----
> From: Baety Wayne SrA 18 CS/SCBX [mailto:Wayne.Baety@kadena.af.mil]
> Sent: Wednesday, 2 October 2002 12:55 PM
> To: 'folivore'; Baety Wayne SrA 18 CS/SCBX; 'Young K. Bae';
> ccielab@groupstudy.com
> Subject: RE: What if OSPF routes have shorter masks than RIP?
>
>
> So far 4 solutions that work...
>
> Yes, what you've stated is called tunnelling the subnet with
> secondaries (1st Solution). That would work in this instance because
> there is only 1 router in his diagram involved on the RIP side. It's
> viable as long as he remembers to place a secondary network throughout
> his RIP domain on each router link (sorta like creating a path) to
> maintain connectivity and reachability information about that
> particular subnet size. (RIP silently drops information about subnets
> across subnets with a differing subnet
size
> but part of the same classful network). It may be better to use a
different
> classful network altogether (2nd Solution) and rely on
> auto-summarization and advertisement of the classful network. He
> could also subdivide the smaller size prefix with secondary networks
> into multiple networks that match the larger subnet size of the RIP
> domain (3rd Solution). To get
this
> into the routing table static addressing would be the the preferred
> method but using secondaries on one of the router links in the OSPF
> domain and relying on proxy-arp to determine destination next-hop
> would also work. Solution 2 and 3 require modification of relatively a
> few routers and probably can be more easily documented than Solution
> 1. A 4th Solution (that was mentioned) is to create a tunneled
> interface matching the subnet size of the RIP domain with the endpoint
> of the tunnel running to each RIP router requiring reachability
> information about that subnet size. This is really just Solution 1
> with performance drawbacks for the sake of convenience and possibly
> smaller subnet utilization (if every router
doesn't
> require reachability information). Both Solution 1 and 4 require
> provisioning a new subnet for each router link just for the sake of
carrying
> information about subnets that have a different subnet size in the
> routing
> table: something that can be quite wasteful if deployed in a large scale
> fashion.
>
> WAYNE A. BAETY, SRA, MCSE, USAF
> 18th Communications Squadron
> DSN: 634-9927
> Commercial: 011 81 611 734 9927
>
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
Donny
> MATEO
> Sent: Tuesday, October 01, 2002 10:04 PM
> To: Richard Davidson
> Cc: ccielab@groupstudy.com; nobody@groupstudy.com
> Subject: RE: What if OSPF routes have shorter masks than RIP?
>
>
> I've tried this and yes it works.
> Care have to be taken though, I was testing on different IOS version
> and guess what...they all got different behaviour. Some works some
> don't. So,
it
> all depends on which IOS you are using.
>
>
> Donny
>
>
>
> Richard Davidson
> <rich@myhomemail. To:
> ccielab@groupstudy.com
> net> cc:
> Sent by: Subject: RE: What if OSPF
> routes have shorter masks than RIP?
> nobody@groupstudy
> .com
>
>
> 28-09-2002 11:11
> Please respond to
> Richard Davidson
>
>
>
>
>
>
> Would a secondary addresses on R2 and R1 in the range of
> 172.16.100.9/29 work? Please let me know if it does. Rich "Young K.
> Bae" wrote:Hi Bob,
>
> The solution stated in the document requires static routes, which I'm
> not allowed to do. Is there any other way to accomplish the object
> without
using
> static routes?
>
> Thanks,
>
> > -----Original Message-----
> > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf
> > Of Bob Sinclair
> > Sent: Friday, September 27, 2002 7:17 PM
> > To: Young K. Bae; ccielab@groupstudy.com
> > Subject: Re: What if OSPF routes have shorter masks than RIP?
> >
> >
> > Try this link:
> >
> > http://www.cisco.com/warp/public/105/52.html
> >
> > It is ugly!
> >
> > ----- Original Message -----
> > From: "Young K. Bae"
> > To:
> > Sent: Friday, September 27, 2002 8:51 PM
> > Subject: What if OSPF routes have shorter masks than RIP?
> >
> >
> > > Guys,
> > >
> > > I'm sure this question has been asked many times before,
> > but can someone
> > > kindly explain to me again? There are some /25 routes in
> > OSPF Area0 that I
> > > need to redistribute into RIP v1. The serial link that
> > connects R1 with R2
> > > has a /26 mask. How can I inject /25 OSPF routes into a
> > classful routing
> > > domain in a case such as this?
> > >
> > > R1 --- 172.16.12.0/26 (RIPv1) --- R2 ---- 172.16.100.0/29
> > (Area1) --- R5
> > > 172.16.200.0/25 (Area0)
> > >
> > > Thanks in advance,
> > >
> > > Young
>
>
>
>
>
> This message is for information purposes only and its content should
> not
be
> construed as an offer, or solicitation of an offer, to buy or sell any
> banking or financial instruments or services and no representation or
> warranty is given in respect of its accuracy, completeness or
> fairness.
The
> material is subject to change without notice. You should take your own
> independent tax, legal and other professional advice in respect of the
> content of this message. This message may contain confidential or
> legally privileged material and may not be copied, redistributed or
> published (in whole or in part) without our prior written consent.
> This email may have been intercepted, partially destroyed, arrive
> late, incomplete or contain viruses and no liability is accepted by
> any member of the Credit Agricole Indosuez group as a result. If you
> are not the intended recipient of this message, please immediately
> notify the sender and delete this message from your computer.
>
>
>
>
>
> This message is for information purposes only and its content should
> not
be
> construed as an offer, or solicitation of an offer, to buy or sell any
> banking or financial instruments or services and no representation or
> warranty is given in respect of its accuracy, completeness or
> fairness.
The
> material is subject to change without notice. You should take your own
> independent tax, legal and other professional advice in respect of the
> content of this message. This message may contain confidential or
> legally privileged material and may not be copied, redistributed or
> published (in whole or in part) without our prior written consent.
> This email may have been intercepted, partially destroyed, arrive
> late, incomplete or contain viruses and no liability is accepted by
> any member of the Credit Agricole Indosuez group as a result. If you
> are not the intended recipient of this message, please immediately
> notify the sender and delete this message from your computer.
This archive was generated by hypermail 2.1.4 : Tue Nov 05 2002 - 08:35:42 GMT-3