Route Summarization for OSPF Multi-Area

daniel280187

Hi Again Guys,

I am currently trying to summarize some Routes in my Area Border Router (ABR).The goal is to be able to summarize the following Networks from my Area 0 so Area 1 Routers can just have Two Routes (One route via CENTR1 and One Route Via CENTR2) instead of 16 routes (Eight routes via CENTR1 and Eight Routes Via CENTR2 - As it's now).



VLAN
Network


Vlan 10
172.16.16.0 / 20


Vlan 20
172.16.32.0 / 20


Vlan 30
172.16.48.0 / 20


Vlan 40
172.16.64.0 / 20


Vlan 50
172.16.80.0 / 20


Vlan 60
172.16.96.0 / 20


Vlan 70
172.16.112.0 / 20


Vlan 80
172.16.128.0 / 20




My Network topology is as follows:




http://imageshack.com/a/img540/140/Xc6YTU.png

Process:I think that the best option to summarize would be 172.16.0.0 / 16. However, this route will include all the networks so I am not quite sure if it is really the most effective way to summarize those networks.My ABR in the Left (CENTR1) and my ABR in the right (CENTR2) are the routers on which I am doing the route summarization.1 - Before the Summarization the Routers in Area 1 (Border 1 - Router in the Left) and (Border 2 - Router in the right) have the following routes in their Routing Tables (Learned via OSPF). I Just want to summarize at this point networks 172.16.0.0 / 16



http://imagizer.imageshack.us/a/img911/1786/dyT4hZ.png


2 - After Configuring Summarization on CENTR 1 (LEFT ABR) the summarisation seems good since the Routers in AREA 1, have now one Route (Summarized Via CENTR 1) and all the other more specific Routes (Not Summarized via CENTR2).CENTR1#show runrouter ospf 1 router-id 10.0.0.0 log-adjacency-changes area 0 range 172.16.0.0 255.255.0.0 area 0 authentication message-digest auto-cost reference-bandwidth 1000000 network 192.168.1.0 0.0.0.255 area 0 network 30.10.1.0 0.0.0.255 area 1



http://imagizer.imageshack.us/a/img673/5167/Wj8jH1.png


3 - After Configuring Summarization on CENTR 2 (RIGHT ABR)router ospf 1 router-id 9.0.0.0 log-adjacency-changes area 0 range 172.16.0.0 255.255.0.0 area 0 authentication message-digest auto-cost reference-bandwidth 1000000 network 192.168.1.0 0.0.0.255 area 0 network 30.10.1.0 0.0.0.255 area 1The problem comes when I configure Summarisation in CENTR 2 (RIGT ABR) since now the Routers in Area 1 Just have one Route to the networks in 172.16.0.0 Via just CENTR1 and not via CENTR2 as they had before. My goal is to load balance the traffic destined to my Vlans via CENTR1 and CENTR2. If a packet was going from Area 1 to area 0 the routers would have two routes to Reach my VLANs 10 -80 Networks.



http://imagizer.imageshack.us/a/img911/4332/X1MWTB.png

Can you notice any mistake or do you have any idea why is this happening, why now I have just one route instead of two?.I have done some debugs to give me a clue but I have not get anything.Thanks for your comments as usual guys!

quickman007

Someone correct me if I'm wrong, but from my understanding the routers are going to pick CENTR-1 or CENTR-2 based on the cost, with the router-id being the tiebreaker if the cost is the same. I assume (can't tell by the pictures) you're using the same cable speeds across the board, so the cost should be equal for either path. the router-id on CENTR-1 is 10.0.0.0 vs 9.0.0.0 on CENTR-2 which would be why you're seeing CENTR-1's route in the table. Again, I could be wrong (more review!).

As for load-balancing on OSPF, I don't think that's taught in the CCNA unless I'm just not up to that section in my studies.

Mooseboost

quickman007 wrote: »

Someone correct me if I'm wrong, but from my understanding the routers are going to pick CENTR-1 or CENTR-2 based on the cost, with the router-id being the tiebreaker if the cost is the same. I assume (can't tell by the pictures) you're using the same cable speeds across the board, so the cost should be equal for either path. the router-id on CENTR-1 is 10.0.0.0 vs 9.0.0.0 on CENTR-2 which would be why you're seeing CENTR-1's route in the table. Again, I could be wrong (more review!).

As for load-balancing on OSPF, I don't think that's taught in the CCNA unless I'm just not up to that section in my studies.

I think equal cost load balancing for OSPF is on the CCNA. I could be wrong but, it was on both of my study materials (max path was part of both OSPF lectures)

daniel280187

quickman007 wrote: »

Someone correct me if I'm wrong, but from my understanding the routers are going to pick CENTR-1 or CENTR-2 based on the cost, with the router-id being the tiebreaker if the cost is the same. I assume (can't tell by the pictures) you're using the same cable speeds across the board, so the cost should be equal for either path. the router-id on CENTR-1 is 10.0.0.0 vs 9.0.0.0 on CENTR-2 which would be why you're seeing CENTR-1's route in the table. Again, I could be wrong (more review!).

As for load-balancing on OSPF, I don't think that's taught in the CCNA unless I'm just not up to that section in my studies.

Hi Quickman, I am not sure whether is part of the CCNA or not, I guess it is since I watched something similar from one of the videos I am using to study for the CCNA.

By default, OSPF does Load Balancing between 4 routes of equal cost. With the command maximum-path (In Router config mode) we can increase or decrease the number of equal cost routes that OSPF can use to load balance.

If you can see the following image http://imagizer.imageshack.us/a/img911/1786/dyT4hZ.png you will find that in each Border Router there are two routes (With equal metric/cost) for each VLAN. As consequence, OSPF will load balance the traffic as you can see in the Routing Table of Each DISTR1 Switch.


My problem comes when I try to summarize the Routes in Area 0 because then the Routers in Area 1 just will see one route Summarised to my VLANs (Via one of the DISTR Routers). What I would like is to have 2 Routes from Each AREA 1 Routers.

OfWolfAndMan

Not sure if this is what you're looking for, but type in the command "show ip route 172.16.0.0". Post the output.

As for your summary, it is what the book would want you to do. However, in a best practice situation, what would you do if you wanted to prevent routing loops from someone who's looking for an IP in that routing statement, but there's actually not a destination?

I.e. 172.16.200.1 isn't actually a part of your configs.

daniel280187

Yes, what I thought is that since the 172.16.0.0 / 16 encompasses all the addresses in that range. If I received a packet from Area 1 destined to a network outside my addressing space, let's say 172.16.200.1 (the IP you suggested) the routers in Area 0 will drop the packets since there is no entry for that network in their Routing Table.

Is there any better way to Summarize those VLANs?



VLAN
Network


Vlan 10
172.16.16.0 / 20


Vlan 20
172.16.32.0 / 20


Vlan 30
172.16.48.0 / 20


Vlan 40
172.16.64.0 / 20


Vlan 50
172.16.80.0 / 20


Vlan 60
172.16.96.0 / 20


Vlan 70
172.16.112.0 / 20


Vlan 80
172.16.128.0 / 20




Or would be better to go back to the IP Planning and maybe choose a better options like this (Summary Route 172.16.0.0 / 21)?



VLAN
Network


Vlan 10
172.16.0.0 / 24


Vlan 20
172.16.1.0 / 24


Vlan 30
172.16.2.0 / 24


Vlan 40
172.16.3.0 / 24


Vlan 50
172.16.4.0 / 24


Vlan 60
172.16.5.0 / 24


Vlan 70
172.16.6.0 / 24


Vlan 80
172.16.7.0 / 24





Another problem I am having right now is that some routes are flapping in my router. Next, you can see the Inputs requested where you will notice the routes Flapping as well. I couldn't notice any information with that command but the part where states "Traffic Share Count is 1" which confirms that the router is not performing any traffic balancing .


BORDER1 - http://imageshack.com/a/img540/7810/cZ6CVh.png
BORDER2 - http://imageshack.com/a/img909/2870/SHjRmD.png


Thanks for your help!

OfWolfAndMan

I assume G0/1 is your outgoing interfaces to your ABRs? If so, I see that the metric on them is quite different. Aside from the reference bandwidth command, what interface configuration to the ABRs do you have between the Border1/2 and the Center1/2?

Also yes, it could drop it, but what if a default route is being advertised? Then you have a potential routing loop. Not one that would bring everything down, but it's an issue for sure. For the fix, you could do something like this:

ip route 172.16.0.0 255.255.0.0 null0
router ospf 1
redistribute static subnets

You put this on both of the routers that have the more specific routes in their RIB (Center1/2). In the case there isn't a more specific prefix, the ABR will drop it. EIGRP has this feature with its summarization built in. OSPF does not.

_Gonzalo_

Have you tried using GNS3?

As you probably know, Packet Tracer has limitations, especially regarding advanced routing. Someone mentioned that routing loops will occur, and that may cause PT to go crazy. This person is right. You should be filtering those (using route maps, for example) to prevent this, but I believe that they do not work well (if at all) on PT.

OfWolfAndMan

_Gonzalo_ wrote: »

Have you tried using GNS3?

As you probably know, Packet Tracer has limitations, especially regarding advanced routing. Someone mentioned that routing loops will occur, and that may cause PT to go crazy. This person is right. You should be filtering those (using route maps, for example) to prevent this, but I believe that they do not work well (if at all) on PT.

Gonzalo does have a point. Get your hands on GNS3 and some IOS .bin files, and if you have the money, I'd recommend Cisco's VIRL over that, if you can figure out configuration by yourself

daniel280187

OfWolfAndMan wrote: »

I assume G0/1 is your outgoing interfaces to your ABRs? If so, I see that the metric on them is quite different. Aside from the reference bandwidth command, what interface configuration to the ABRs do you have between the Border1/2 and the Center1/2?

Also yes, it could drop it, but what if a default route is being advertised? Then you have a potential routing loop. Not one that would bring everything down, but it's an issue for sure. For the fix, you could do something like this:

ip route 172.16.0.0 255.255.0.0 null0
router ospf 1
redistribute static subnets

You put this on both of the routers that have the more specific routes in their RIB (Center1/2). In the case there isn't a more specific prefix, the ABR will drop it. EIGRP has this feature with its summarization built in. OSPF does not.

Hi OfWolfAndMan,

I assume G0/1 is your outgoing interfaces to your ABRs? - Yes, that is Right Gi 0/1 and Gig 0/2 are my Interfaces from my Area 1 Routers to my ABR.

what interface configuration to the ABRs do you have between the Border1/2 and the Center1/2?

BORDER 1
interface GigabitEthernet0/1
ip address 30.10.1.1 255.255.255.252
duplex full
speed 1000
!
interface GigabitEthernet0/2
ip address 30.10.1.9 255.255.255.252
duplex full
speed 1000


CENTR1 (ABR)

interface GigabitEthernet0/1
no switchport
ip address 30.10.1.2 255.255.255.252
duplex full
speed 1000
!
interface GigabitEthernet0/2
no switchport
ip address 30.10.1.6 255.255.255.252
duplex full
speed 1000





BORDER 2

interface GigabitEthernet0/1
ip address 30.10.1.13 255.255.255.252
duplex full
speed 1000
!
interface GigabitEthernet0/2
ip address 30.10.1.5 255.255.255.252
duplex full
speed 1000


CENTR 2
interface GigabitEthernet0/1
no switchport
ip address 30.10.1.14 255.255.255.252
duplex full
speed 1000
!
interface GigabitEthernet0/2
no switchport
ip address 30.10.1.10 255.255.255.252
duplex full
speed 1000


Thanks to your suggestion I spotted a mismatch on some routers for the auto-cost reference-bandwidth 1000000 since some of them were configured for 100 M and some others for 1,000 M so now is all matching. Additionally I hard coded all the Gigab interfaces to work at 1 Gbps and in Full Duplex mode which eventually solved the problem of the routes/interfaces Flapping. However, I am still not able to balance the traffic from my Area 1 to my ABRs via the two Gig Interface (Gig 0/1 and Gig 0/2).


Luckily I have not configure any default-route or had honestly think about that scenario where a Routing loop would decrease the performance of the network, but it really makes sense so thanks for the suggestion

ip route 172.16.0.0 255.255.0.0 null0
router ospf 1
redistribute static subnets

I don't have a clue of this configuration so I will get more information on the Internet on how can I use it and adapt it for my current setup.

I totally Agree with you and Gonzalo, nothing would be better than trying this configs on real gear or at least in GNS3 but I can't right now so the best tool I've got available is Packet Tracer.

I hadn't hear about that VIRL so I was looking some videos and seems pretty interesting and even more powerful than GNS3.

Thanks for the help guys!

_Gonzalo_

daniel280187 wrote: »

ip route 172.16.0.0 255.255.0.0 null0
router ospf 1
redistribute static subnets

I don't have a clue of this configuration so I will get more information on the Internet on how can I use it and adapt it for my current setup.

I totally Agree with you and Gonzalo, nothing would be better than trying this configs on real gear or at least in GNS3 but I can't right now so the best tool I've got available is Packet Tracer.

No need to look it up, it´s quite simple. But remember this is CCNP stuff...

Redistribute will add routes to your dynamic routing protocol. In this case, it adds all static routes to ospf. Redistribution is slightly different for every routing protocol. OSPF will only redistribute classful networks unless you add "subnets" at the end. In the case of RIP or EIGRP, you need to set the metric for routes to be redistributed.

I understand you not wanting to switch from PT to GNS3 (PT is so cool!!) but you will eventually have to... This is free software, so money shouldn´t be an issue. And you really do not want to be guessing when PT is being the issue or you config is wrong. And that is not the only issue... PT tends to crash when your network is too complicated, and you are one step from that!

Trust me, you want to go for GNS3 as soon as you can!

daniel280187

Thanks for the comments Gonzalo, just now I was able to read the forum again.

I had 200-120 exam yesterday and passed .

I really wanted to switch to GNS3 (And eventually work with real gear) but since my exam date was that close did not want to waste any time learning how to set up a basic lab. Now that I have got time will start learning on how to use GNS3 to set up new labs along with Wireshark that seems an excellent way to really and deeply understand how the packets are routed and what is actually inside those packets.

I am really grateful for your help guys.

OfWolfAndMan

Congrats, Dan! Good stuff. I can say Wireshark is definitely a great method of learning packet types.

_Gonzalo_

Cool man!! Congrats!