About Discontiguos Network

[kumaran]

Hi,
I'm confused about the discontiguous network. Can anyone help me in this..
please
Thanks

[aguilera]

Have you ever eaten at a buffet? Well I have. I have been to plenty of them. We are getting to the discontiguous part... just give me a second to get there.

At a buffet, everyone grabs a plate and lines up and starts loading it on! The buffet is usually set up with similar entrees together, then perhaps the side items, salads, breads then the good part... the deserts! :confused:
We're getting to the discontiguous part, I promise.

Ok, lets say that you are at a buffet and you want pasta. You look in pasta area and there are only two pastas to choose from. You want penne, but the only thing you can find is linguini, and spaghetti. You then set your plate down, get a refund and leave. Why? Because you wanted penne and the only thing you could see was linguini and spaghetti.

The bad thing about this incident is that there was penne pasta.

Unfortunately all the pasta items were not contiguous, they were not all put together. The pene pasta had been separated from all the other pastas by the salad section. If only you were able look beyond the salad section you would have realized that your penne pasta was there.

This is the case with discontiguous networks.

Classfull routing protocols will not let you look beyond the "salad"

Classless routing protocols will allow you to peer beyond the uncommon network that might be creating a separation between the pasta. How? They send mask information with their routing updates.

Look in my next post. There I will use an example using IP addresses versus Pasta :D

[aguilera]

Classfull network Example:
..................___ ...................___ .....................___.............
10.10.10.0 | A |172.16.10.0 | B | 192.168.10.0 | C | 10.1.1.0

A's Routing Table:
10.10.10/24 Why 24? Because it is directly connected
172.16.10/24 Why 24? Because it is directly connected
192.168.10/24 Why 24? 192 is a class C with a default mask of /24

B's Routing Table:
10.0.0.0/8 outboud towards Router A
10.0.0.0/8 outboud towards Router B Why 2 routes? Equal cost! (Bad Move)
172.16.10/24 Why 24? Because it is directly connected
192.168.10/24 Why 24? 192 is a class C with a default mask of /24

C's Routing Table:
10.1.1/24 Why 24? Because it is directly connected
172.16/16 Why 16? Because it is a class B with a default mask of /16
192.168.10/24 Why 24? 192 is a class C with a default mask of /24

Why doesn't Router A or Router C document each others 10 network?

Well the fact is Route A is sending an up date to Router B and in turn Router B is sending an update to Router C, Router C is just not using it. Why?
Router C says I already have a route to the 10.0.0.0 and it is directly connected... so why should I place a route in my routing table that will cost me more hops to get to than I already have.

Let's take a look. What does Router A's routing update look like?
Router A is telling Router B it can get to network 10.0.0.0 and it will only cost Router B 1 hop.

Router B then sends that information on to router C. Router B tells Router C it can get to the 10.0.0.0 network and it will only cost Router C 2 hops.

Well when router C sees that, Router C kinda chuckles to it self and says "I already have a route to the 10.0.0.0 in my routing table and it is less hops that this one...thanks but no thanks!

The same thing would happen when Router C pushes routes to Router B and then from B to A.

How do we fix it? Using Routing protocols that will carry mask information.

Part 3 of this post will have that information....

[aguilera]

Classless network Example:
..................___ ...................___ .....................___.............
10.10.10.0 | A |172.16.10.0 | B | 192.168.10.0 | C | 10.1.1.0

A's Routing Table:
10.10.10/24 Why 24? Because it is directly connected
10.1.1.0/24 outboud towards Router B (Routing protocol sends Mask info)
172.16.10/24 Why 24? Because it is directly connected
192.168.10/24 Why 24? 192 is a class C with a default mask of /24

B's Routing Table:
10.10.10.0/24 outboud towards Router A (Routing protocol sends Mask info)
10.1.1.0/24 outboud towards Router B (Routing protocol sends Mask info)
172.16.10/24 Why 24? Because it is directly connected
192.168.10/24 Why 24? 192 is a class C with a default mask of /24

C's Routing Table:
10.10.10.0/24 outboud towards Router B (Routing protocol sends Mask info)
10.1.1/24 Why 24? Because it is directly connected
172.16.10/24 Why 24? (Routing protocol sends mask information)
192.168.10/24 Why 24? 192 is a class C with a default mask of /24

You notice now that both Router A and Router C have classess routes? Even though they are separated with uncommon subnets! How was this possible?

Let's take a look at what Router A's routing update looks like.

Router A is telling Router B it can get to network 10.10.10/24 and it will only cost Router B 1 hop.

Router B then sends that information on to router C. Router B tells Router C it can get to the 10.10.10.0 network and it will only cost Router C 2 hops.

Well when router C sees that, Router C now says hey I do have a route to the 10.1.1/24 network, but I do not have a route to the 10.10.10/24 network, so let me put that in my routing table. Router A will end up doing the same thing when it gets a routing update from Router B.

The big difference is that now the routing protocol is being specific about what is being advertised.

Classless routing helps route traffic in discontiguous networks by sending more detailed mask information...the ability to look beyond the salad.;)

[kumaran]

Thanks sir,
I'm really in confused state about Discontiguous Network...But your reply cleared my doubts
Once again Thanks alot!!!

[aguilera]

Glad the post was able to help out!

[zjm]

been looking for a good explanation of discontigous networks and have to see that explained it all very well. Thanks!

[lildeezul]

Ok, i think i get it.
So, using contigious networks, like classfull networks, router A only send the 10.0.0.0 update to router B, and router B tells router C about the 10.0.0.0 network, but router C says i already have this network and its directly connected,
but in reality, router C has the 10.1.1.0 network, and Router A has a 10.10.10.0 network, but since a classfull protocol was used Router A, and Router C will only send to classfull network such as 10.0.0.0, right???

if were using OSPF that supports discontigious networks by defualt. then Router A running OSPF will send updates to router B sayin " i know about network 10.10.10.0"
as oposed to 10.0.0.0

Am i correct.??? i belive i understand this.
Also for use of support of discontigous network in EIGRP, then i would have to issue the no summary command. Correct.?????

contigious network dont work in class C address do they??

only in class B and A am i correct??

[mithra]

Hi,
I am unable to understand the summarization.. I am referring Todd's 6th edition...
could you plz explain me little more clearly
Thanks
Mithra

[mithra]

Hi Lammle,
Plz explain the discontiguous network .. though i ref the previous posts by Aguilera still unable to follow it
thanx
Mithra

[lildeezul]

ok let say we have three routers..

Router A, Router B, and Router C

Imagine Router A was connected to Router B via serial, and Router B connected to Router C via Serial,

also imagine that every router has a ethernet port with a broadcast multi-access network connected to it.

Router A--------------------Router B------------Router C

If all routers were using RIP ( a classful routing protocol)

imagine that Router A ethernet port had a network with an ip address space of 10.1.1.0/24

and Router C ethernet port had a network with with an ipp address space of 10.2.2.0 /24

RIP uses only classful address..

So on routers B routing table it would be similar to this.

R ------- 10.0.0.0. (router A)
R--------10.0.0.0 (router B)_

this is a contiguous network becuase the routing protocol summarized both network from Router C, and router A as 10.0.0.0..

so how do router B know which Router to send data to if it has two routes to the 10.0.0.0 network ???


hope you get it

[lammle]

1. you need to set up these three routers and try this yourself.
2. RIPv1 will find the subnets and advertise them to each routers, as long as:
A. All networks used the same subnet mask
B. every router interface had a 10.0.0.0 network.

If you can't answer A and B, then you need to use RIPv2.

Todd

[sukarabi]

Hi everybody,

I tried to setup Aguilera's (thanks for your pasta explanation who made me more confortable with discontiguous net) example but in Todd's Hands-on-lab6 (chapter 6 page 403), then I was able to make the lab and test DN at the same time.
In fact I added 10.10.10.1/24 interface on LabA and 10.1.1.1/24 interface on LabC
Activate 10.0.0.0 in RIP protocol on LabA and labC
check my routing tables and start debug on each router (A,B and C).
So it's like Aguilera said for the routing tables:
on LabA I can see just 10.10.10.0 network directly connected
on labB 10.0.0.0 is referenced twice by Rip (on each interface)
on labC just 10.1.1.0 network directly connected
until here nothing strange (except if I did a mistake)
The problem is when I look at the debug results, my LabC doesn't receive any advertisement about any 10.0.0.0 on his serial interface (connected to labB)
Same for LabA, nothing about any 10.0.0.0 on his serial interface (connected to labB too but on the other side).
It's like labB doesn't propagate the rumor he received on both side.
Looking at labB debug confirmed that it doesn't send advertisement for these networks.
I didn't see any advertisement of 10.0.0.0 with more hops than the directly connected one.

So did I forget something in my config ?
Thanks for any help.

[sukarabi]

Ok no need to investigate more I found a solution to my problem.
As I was doing the hands-onlab-7 (eigrp one) I decided to go further and add some discontiguous network on the edges of my eigrp network.
So the result, for sure, was the same as with RIP.
But with EIGRP, because I had the option to use the no auto-summary command I tryied to make it working.
After using "no auto-summary" on the edge routers, no change happened.
I was still not able to see the opposite one in the routing tables.
The Hands-on-lab-7 is composed of 4 routers, so on the left 2 routers I was just able to see the left discontiguous network. Idem on the 2 right ones with just the right discontiguous network.
The solution came after I used "no auto-summary" on all of the routers of the line.
Anyone has a good explanation for this problem ? Is that something to do with Split Horizon for the router in the middle with auto-summary on ?
Thanks for any help.

[lildeezul]

The solution came after I used "no auto-summary" on all of the routers of the line.
Anyone has a good explanation for this problem ? Is that something to do with Split Horizon for the router in the middle with auto-summary on ?
Thanks for any help.

The issue is when one router recieves updates from other router and have a classful address to them.
and then That same router recieves other routes that are VLSM supported.

you need to have all the routers configured with the no auto-summ command, because if not then the router will recieve some routes that are auto summarized to the classful boundary, and other routes that are more specific. This could cuase loss of packets and loops in a network,