Reason of why all areas must connect to the backbone area?

Well in OSPF is like this and im wondering why

Most of the stuff in OSPF works like a link state routing protocol... at least for example inside an area it works like this
Routers inside an area exchange type 1 and type 2 LSA to build the topology table(LSDB) them use the SPF algorithm agains it to do the routing table.
Then when another router recive a network summery LSA from an ABR it does not run the SPF algorith it simple add it to the routing table and it uses the cost to that destination from the ABR so this part works as a distance vector protocol...

Now i was reading some books and it says this is the reason of why all areas must connect to the backbone area.... cause of this distance vector behavior it says it can avoid loops
I just dont see how this works... andwhy it avoid loops... and i ve been trying to find why but all i see is that they mention it but NEVER EXPLAIN it properly with dunno maybe an example in the documents or books i have read it just says that it must connect to the area 0... sigh

im confused someone help me?

Find more posts tagged with

Save $250 on 2025 certification boot camps from Infosec!

Book now with code EOY2025

Button

Comments

rossonieri#1

hello,

avoiding loops in the context of that OSPF must be well-designed using hierarchical ip-subnet. the highest IP gets the preference.

cheers.

NightShade1

I dont think you got my question..... because that doesnt asnwer it but thanks for the reply anyways

Harryhcl

The answer is because OSPF uses the inter area and intra area to identify routes. A routers in any area that is not area 0 will have a copy of the database for that area unless it is an ABR. The ABR has to have two databases (or one hierarchical one), one for the area and one for area 0. If there was no area 0 all routers would have to have very big and flat databases. So it uses area 0 as a conveinient passing place for inter area routes.

Not a very scientific anser I know but can it keep you sane?

rossonieri#1

well, AFAIK i think that is the answer for the question.

from the post the ABR acting like distance-vector router? there is no such a thing as distance-vector routing in OSPF - its not a hybrid routing protocol like EIGRP. pls read the requirement for OSPF implementation.

cheers..

tech-airman

NightShade1 wrote:

Well in OSPF is like this and im wondering why

Most of the stuff in OSPF works like a link state routing protocol... at least for example inside an area it works like this
Routers inside an area exchange type 1 and type 2 LSA to build the topology table(LSDB) them use the SPF algorithm agains it to do the routing table.
Then when another router recive a network summery LSA from an ABR it does not run the SPF algorith it simple add it to the routing table and it uses the cost to that destination from the ABR so this part works as a distance vector protocol...

Now i was reading some books and it says this is the reason of why all areas must connect to the backbone area.... cause of this distance vector behavior it says it can avoid loops
I just dont see how this works... andwhy it avoid loops... and i ve been trying to find why but all i see is that they mention it but NEVER EXPLAIN it properly with dunno maybe an example in the documents or books i have read it just says that it must connect to the area 0... sigh

im confused someone help me?

NightShade1,

I was certified under the CCNA and CCDA programs back in 2003. I was studying for the BSCI in order to attempt to recertify my CCNA and CCDA but didn't take and pass the BSCI exam in time to recertify my Associate level certificates. I am now currently studying for the 640-801 CCNA exam to recertify from scratch. Disclaimer: Since I am studying for the 640-801 CCNA exam, I may venture into unknown areas (pun intended). :P

Part of the current 640-801 CCNA exam is to configure OSPF in a single area. Well, I'm learning that "single area" refers to the backbone or area 0. As you may or may not know, each interface in an OSPF router belongs to an area. If all interfaces of an OSPF router belong to a single area, like area 0, then the router is called an "Internal Router." As you may or may not recall, if there are multiple internal routers, then they participate in the election of the Designated Router (DR) and the Backup Designated Router (BDR). The Designated Router for the area knows the routes for all the networks within the area. The DR for area 0 is special, but I'll explain that later.

Now, let's say you want to add a stub area, say area 1. Once again, if all the interfaces of a router belong to the same area, in this case area 1, then that router is also an internal router for area 1. Then assume there's enough routers to participate in the DR election process. Eventually, a DR for area 1 is elected. Now, how do you route packets from area 1 to area 0 ? That's when you use an "Area Border Router" or ABR for short. The ABR has at least two interfaces, one that belongs to area 1, and one that belongs to area 0. The DR of area 1 sends routing information to the ABR, then the ABR relays the area 1 routing information to the DR of area 0. Now, the DR for area 0 knows the routing information for both area 0 and area 1.

Now, let's say you want to add another stub area, say area 2. Let's assume there's enough internal routers for area 2 to elect a DR. Let's also assume there's an ABR between area 0 and area 2. So the DR for area 2 sends routing information to the ABR, then the ABR relays the area 2 routing information to the DR of area 0. Now the DR for area 0 knows all the routes for area 0, area 1, and area 2. This is what makes area 0 important for OSPF. After the DR for area 0 knows the routes for area 0, area 1, and area 2, now packets from area 1 can now be routed to area 2 via area 0. That's why area 0 is called the backbone.

So to answer your "loop avoidance" question, it avoids loops because:

A packet is sent from a host in area 1 towards a host in area 2. The packet is first forwarded to the host's default gateway router in area 1.
Since the network for area 2 is unknown by the internal routers of area 1, the router in area 1 forwards the packet to the ABR between area 1 and area 0.
The ABR between area 1 and area 0 forwards the packet to the ABR between area 0 and area 2.
The ABR between area 0 and area 2 forwards the packet to the router that is connected to the network where the host in area 2 is connected to.
The area 2 internal router forwards the packet from itself to the host.

So therefore, a routing loop is avoided because area 0, the backbone, is the default network where packets with any unknown destination network address within the stub area are sent and is fowarded appropriately through the backbone to reach the destination network in another area.

Maybe the reason why you may feel that OSPF "seems like a distance vector" is how packets go from router to router in a "hopping" fashion. However, what makes OSPF NOT a distance vector routing protocol is because SPF is used to create the routing table, not a distance vector's routing protocol, like RIP's "hop count" routing table entries.

I hope this helps. Since I'm not studying for the 642-801 BSCI anymore, I do not know the technical steps to configure OSPF in multiple areas. Since I disclaimed earlier, anyone who knows better may correct anything I've said.

Source:

OSPF Design Guide - Cisco Systems - http://www.cisco.com/en/US/tech/tk365/technologies_white_paper09186a0080094e9e.shtml
How does OSPF Generate Default Routes? - Cisco Systems - http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094a74.shtml#typesofospfareas

NightShade1

rossonieri#1 wrote:

well, AFAIK i think that is the answer for the question.

from the post the ABR acting like distance-vector router? there is no such a thing as distance-vector routing in OSPF - its not a hybrid routing protocol like EIGRP. pls read the requirement for OSPF implementation.

cheers..

rolf rossonieri#1 the one that should be reading books itsYOU because thats is how it works.... i highly recomend you routing tcp/ip of jeff doyle

Yeah it doesnt say it in most of books i ve read because its advance knowledge... go and ask a CCIE about it if you dont bealive me... I know it because i like reading some advance books but sometimes i dont get stuff i read just like the one im asking

Umm now lets start reading asnwers of the other ppl

NightShade1

tech-airman wrote:

NightShade1 wrote:

Well in OSPF is like this and im wondering why

Most of the stuff in OSPF works like a link state routing protocol... at least for example inside an area it works like this
Routers inside an area exchange type 1 and type 2 LSA to build the topology table(LSDB) them use the SPF algorithm agains it to do the routing table.
Then when another router recive a network summery LSA from an ABR it does not run the SPF algorith it simple add it to the routing table and it uses the cost to that destination from the ABR so this part works as a distance vector protocol...

Now i was reading some books and it says this is the reason of why all areas must connect to the backbone area.... cause of this distance vector behavior it says it can avoid loops
I just dont see how this works... andwhy it avoid loops... and i ve been trying to find why but all i see is that they mention it but NEVER EXPLAIN it properly with dunno maybe an example in the documents or books i have read it just says that it must connect to the area 0... sigh

im confused someone help me?

NightShade1,

I was certified under the CCNA and CCDA programs back in 2003. I was studying for the BSCI in order to attempt to recertify my CCNA and CCDA but didn't take and pass the BSCI exam in time to recertify my Associate level certificates. I am now currently studying for the 640-801 CCNA exam to recertify from scratch. Disclaimer: Since I am studying for the 640-801 CCNA exam, I may venture into unknown areas (pun intended). :P

Part of the current 640-801 CCNA exam is to configure OSPF in a single area. Well, I'm learning that "single area" refers to the backbone or area 0. As you may or may not know, each interface in an OSPF router belongs to an area. If all interfaces of an OSPF router belong to a single area, like area 0, then the router is called an "Internal Router." As you may or may not recall, if there are multiple internal routers, then they participate in the election of the Designated Router (DR) and the Backup Designated Router (BDR). The Designated Router for the area knows the routes for all the networks within the area. The DR for area 0 is special, but I'll explain that later.

Now, let's say you want to add a stub area, say area 1. Once again, if all the interfaces of a router belong to the same area, in this case area 1, then that router is also an internal router for area 1. Then assume there's enough routers to participate in the DR election process. Eventually, a DR for area 1 is elected. Now, how do you route packets from area 1 to area 0 ? That's when you use an "Area Border Router" or ABR for short. The ABR has at least two interfaces, one that belongs to area 1, and one that belongs to area 0. The DR of area 1 sends routing information to the ABR, then the ABR relays the area 1 routing information to the DR of area 0. Now, the DR for area 0 knows the routing information for both area 0 and area 1.

Now, let's say you want to add another stub area, say area 2. Let's assume there's enough internal routers for area 2 to elect a DR. Let's also assume there's an ABR between area 0 and area 2. So the DR for area 2 sends routing information to the ABR, then the ABR relays the area 2 routing information to the DR of area 0. Now the DR for area 0 knows all the routes for area 0, area 1, and area 2. This is what makes area 0 important for OSPF. After the DR for area 0 knows the routes for area 0, area 1, and area 2, now packets from area 1 can now be routed to area 2 via area 0. That's why area 0 is called the backbone.

So to answer your "loop avoidance" question, it avoids loops because:
A packet is sent from a host in area 1 towards a host in area 2. The packet is first forwarded to the host's default gateway router in area 1.

Since the network for area 2 is unknown by the internal routers of area 1, the router in area 1 forwards the packet to the ABR between area 1 and area 0.

The ABR between area 1 and area 0 forwards the packet to the ABR between area 0 and area 2.

The ABR between area 0 and area 2 forwards the packet to the router that is connected to the network where the host in area 2 is connected to.

The area 2 internal router forwards the packet from itself to the host.

So therefore, a routing loop is avoided because area 0, the backbone, is the default network where packets with any unknown destination network address within the stub area are sent and is fowarded appropriately through the backbone to reach the destination network in another area.

Maybe the reason why you may feel that OSPF "seems like a distance vector" is how packets go from router to router in a "hopping" fashion. However, what makes OSPF NOT a distance vector routing protocol is because SPF is used to create the routing table, not a distance vector's routing protocol, like RIP's "hop count" routing table entries.

I hope this helps. Since I'm not studying for the 642-801 BSCI anymore, I do not know the technical steps to configure OSPF in multiple areas. Since I disclaimed earlier, anyone who knows better may correct anything I've said.

Source:
OSPF Design Guide - Cisco Systems - http://www.cisco.com/en/US/tech/tk365/technologies_white_paper09186a0080094e9e.shtml

How does OSPF Generate Default Routes? - Cisco Systems - http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094a74.shtml#typesofospfareas

Lol i dont feel it like it, They got a distance verctor behavior and it is like that.... and let me correct you some stuff
For example the SPF just happens with LSA typ1 and type 2 but for LSA type 3 which is the summery route that the ABR inject to to other areas it doesnt do the SPF it just add it to the topology table and routing table with the cost to the ABR

It is a vector distance behavior because of this.... because LSA typ 3 doesnt use the SPF to add it to the routing table it JUST ADD IT but it doesnt use the SPF like with the LSA type 1 and 2.... the info that got LSA type 3 include also the cost to the abr

So in summery what i mean with all this?
Well OSPF is a link state protocol inside an area but It uses a distance vector algorithm to find interarea routes this is the key(this is the part that you dont know rossonieri#1)

btw Harryhcl let me see if i got you
Routers in other area have a copy area 0 database??? O_o
i guess got your point but im not that sure... would be too much asking if you give me a small example harry?

ummm ill simulate it in dynamips and start checking the databases and routing tables again...

rossonieri#1

So in summery what i mean with all this?
Well OSPF is a link state protocol inside an area but It uses a distance vector algorithm to find interarea routes this is the key(this is the part that you dont know rossonieri#1)

well,

as you wish then --
tell me which part of OSPF uses the hop count?
AFAIK,
OSPF calculate its own metric -
OSPF will know nothing about other area unless you tell your router to do so.

cheers.

optimus

OSPF is by no means a distance vector protocol.

It is link state. I am not sure where you are getting that, "a distance vector algorithm to find interarea routes ", but if you tell me your source, I will look at it myself and try and interpret it. LSA (Link State Advertisements) are sent out when there are changes to the topology and those are sent to the DR or BDR. The LSAs let the routers create the topological database for themselves for their own area (backbone is Area 0), and as others stated ABRs get two databases (not sure if they can get more than two). The DIJKSTRA ALGORITHM is used on every router's topological database to get the shortest path to each of the networks, and the routing table is built for that router. It is that topological database and the DIJKSTRA ALOGORITHM that prevents the loops from occuring. The algorithm goes into that database and says, "HERE IS THE PATH YOU USE". It surely won't return a looped path. That is the beauty of the algorithm.

I hope this helps. Let me know if it doesn't and we can converse about this further.

EdTheLad

I'm not sure if the question was answered, but i'll fire away anyway.
Ospf uses type 1 and 2 lsas to create a database and from this database the spt is calculated,the resultant shortest paths a each destination is put in the routing table.
Ospf floods type 3 and 5 lsas into an the type 3 lsas will have an originator id of the ABR which all routers in the area wil already know, the type 5 lsas will have an originator id of the asbr which all internal routers will not know.So type 4 lsas are also flooded to tell all internal routers the cost to get to the asbr.
Just looking at type 3 lsas,you can see they are not used to create the database, the route information is just added to the result of the spf tree.These type 3 lsas are flooded throughout the area and known to be created by the abr.Since these lsas are flooded and not used in the database calculation, these are effective using routing by romour, where is abr is the source of the rumor, distance vector like.Since each area has a direct connection to where the rumor originated i.e. the backbone its not possible for this route info to create a loop.If you had multiple areas interconnected without the backbone rule, and lsas were being flooded, a single area might learn the same route from different areas.Since there is no split horizon or holddown timers between areas you can have a loop.So its best to view each area like a distance vector router, using the backbone as the single route source stops multiple paths to a destination and loops from being formed.
This distance vector routing behavior is also seen with bgp, bgp uses AS_path to counteract the behavior.Its a shame they dont use the same idea in ospf,an area_path could be added to a route update and there would be no need for the backbone rule but im sure they have their reasons.

NightShade1

EdTheLad you owner!!!

thank you, you were the only one th at aswered my question hehe (well maybe harry also did but i dindt understand his explanation)

And well you also explained all the process now i don thave to explain it to rossonieri#1 well i ll actually explain a bit)

"Since there is no split horizon or holddown timers between areas you can have a loop"
This was the KEY sentence... i wasnt thinking that part.....................
you did really enlight me man

"using the backbone as the single route source stops multiple paths to a destination and loops from being formed. "
So it works like a hub and spoke topology doesnt it??

"OSPF calculate its own metric"
I alredy explained it on my preview post which you arent reading carefully or my english just blows too much

It just use LSA type1 and type 2 to build the database then do the SPF gains the LSDB to build the routing table... then LSA type 3 is flooded from the ABR and it just added to the table i mean in simple words like EdTheLad said its just add it to the result of the SPF calculation so it doesnt use the SPF to calculate interarea routes!!!!

Now the LSA typ3 includes the cost to the ABR.... now you will ask me and which cost is using if it doesnt use any cost calculated by the SPF algorithm? well it use the interface outgoing cost for it, as far i remneber.... you can google about it.

"tell me which part of OSPF uses the hop count? "
Do you actually know what distance vector is? it has nothing to do with the hop count...
Distance verctor as the name implies means that the routes are advertised as verctors of distance, direction..... where the distance is in term of metric(which can be hop count like in rip or can be like in IGRP that uses delay, bandwidth, reliability, and load to calcuate the metric) and direction is the next hop he should go to reach the destiny

OSPF is not using hop count yeah... what i mean there is that is using routing by rumor being the ABR the source of the romour... which mean that the router are depending on others for the information in this inter area part. So this mean its using distance vector algorithm to find interarea routes

"OSPF will know nothing about other area unless you tell your router to do so. "

The routers inside an area knows about the other areas because of the ABR which is the source of the routing rumor..... (inject the LSA type3 in the area and it add it to the info that was alredy calculated by the spf)

Anyways you can read everything about this process in EdTheLad post, he did a quite good resume of it.... its pretty good

Ah yeah optimus my source is this book "Routing TCP/IP for CCIE professionals by jeff doyle" I love this book heh
You should read also more as well optimus...

PD: what means AFAIK???

optimus

Ah good, I am glad that Ed was able to correctly answer your question. Sorry that my reply was not sufficient.

Yeah, that text by Jeff Doyle is a classic text. I should probably pick myself up a copy.

NightShade1

Yes you should buy it... the book is awsome.

rossonieri#1

not all,

"Since there is no split horizon or holddown timers between areas you can have a loop"
what makes a routing loop? and why there is no routing loop in OSPF?

"Do you actually know what distance vector is? it has nothing to do with the hop count.."
actually the first thing about distance vector is about the hop count.

"The routers inside an area knows about the other areas because of the ABR which is the source of the routing rumor..... (inject the LSA type3 in the area and it add it to the info that was alredy calculated by the spf)"

tell me this diagram :

area 1 --> area 0 <-- area 1 <--area 2 <--area 3

does the most left area 1 knows about area 3?
and, which part of it act like distance vector to know the area 3?

cheers.

NightShade1

Okay lets start hehe
Firts
area 1 --> area 0 <-- area 1 <--area 2 <--area 3

This diagram is not valid... all areas must connect to area 0.....
Area 3 and area 2 arent connected to Area 0.... guess you could connect area 2 to area 0 with a virtual link O_o
this one is valid
area 3 --> area 0 <-- area 1
Look its a distance vector just because it using routing by roumour i mean the ABR of area 0 and area 1 its telling the routers inside Area 1 that he can comunicate them to other areas for example area 3.... for this he is NOT using the routing table it got from doing the SPF algorithm over the LSDB... i mean routes to other areas are not part of this calculation when ABR injects LSA type 3 into the area it just ADD them to the table... it dosent make any calculation....

In simple words the behavior of depending on an intermediate router instead of determining the full route to the destination (this is the key sentence) is a distance vector behavior!!!!!!!
This is what i want you to see its DEPENDING of a intermediate router is not determining the full route to the destination calculating it with the SPF!!!!! i mean the inter area routes!

"what makes a routing loop? and why there is no routing loop in OSPF? "
There is no loop because all the areas are connected to Area 0!
That is what i was asking i mean why all the areas must connect to area 0! i knew the rule but didnt knew why, and this is what i was asking... if you wanna know more im not typing it Again and again over and over... just read my post or better! read EdTheLad post its all explained there pretty good...

"actually the first thing about distance vector is about the hop count. "

No you are confused Look i alredy explained it.....

Vecotor=direction which means the next router he has to go
Distance=METRIC METRIC ok?? the metric can be hop count like in rip or can be like in IGRP which use delay, bandwidth, reliability, and load to calcuate the metric
this is not about the HOP count!!! jeez!

The difference between distance vector and link state is that the distance vector DEPENDE of other routers! link state doesnt depend! because it independently calculate its own best path!!!! because its has its own network map unlike distance vector which just know what his neighbors knows!

Look OSPF is link state inside the ARea but for interarea routeing it use vector distance
Why you dont see this in some books ? well i dont ****** know.... at least you woudlnt see that in CCNA because they just focus in OSPF in Area 0....

you got it now?
Fallow your own advice and go and read a book about it....

PD: You havent told me what means AFAIK

Pash

AFAIK - As far as I know - seeing as you seem quite resourceful Nightshade im surprised you havent googled that yet

Waiting for my BSCI book to come in the post this week, can't wait

)

tech-airman

NightShade1 wrote:

Okay lets start hehe
Firts
area 1 --> area 0 <-- area 1 <--area 2 <--area 3

This diagram is not valid... all areas must connect to area 0.....
Area 3 and area 2 arent connected to Area 0.... guess you could connect area 2 to area 0 with a virtual link O_o
this one is valid
area 3 --> area 0 <-- area 1
Look its a distance vector just because it using routing by roumour i mean the ABR of area 0 and area 1 its telling the routers inside Area 1 that he can comunicate them to other areas for example area 3.... for this he is NOT using the routing table it got from doing the SPF algorithm over the LSDB... i mean routes to other areas are not part of this calculation when ABR injects LSA type 3 into the area it just ADD them to the table... it dosent make any calculation....

According to RFC 1583 - OSPF Version 2, "OSPF is a link-state routing protocol."

Source:

http://www.ietf.org/rfc/rfc1583.txt?number=1583 - http://www.ietf.org/rfc/rfc1583.txt?number=1583

NightShade1 wrote:

In simple words the behavior of depending on an intermediate router instead of determining the full route to the destination (this is the key sentence) is a distance vector behavior!!!!!!!
This is what i want you to see its DEPENDING of a intermediate router is not determining the full route to the destination calculating it with the SPF!!!!! i mean the inter area routes!

According to RFC 1583 - OSPF Version 2, "OSPF is a link-state routing protocol."

Source:

http://www.ietf.org/rfc/rfc1583.txt?number=1583 - http://www.ietf.org/rfc/rfc1583.txt?number=1583

NightShade1 wrote:

"what makes a routing loop? and why there is no routing loop in OSPF? "
There is no loop because all the areas are connected to Area 0!
That is what i was asking i mean why all the areas must connect to area 0! i knew the rule but didnt knew why, and this is what i was asking... if you wanna know more im not typing it Again and again over and over... just read my post or better! read EdTheLad post its all explained there pretty good...

"actually the first thing about distance vector is about the hop count. "

No you are confused Look i alredy explained it.....

Vecotor=direction which means the next router he has to go
Distance=METRIC METRIC ok?? the metric can be hop count like in rip or can be like in IGRP which use delay, bandwidth, reliability, and load to calcuate the metric
this is not about the HOP count!!! jeez!

The difference between distance vector and link state is that the distance vector DEPENDE of other routers! link state doesnt depend! because it independently calculate its own best path!!!! because its has its own network map unlike distance vector which just know what his neighbors knows!

OSPF depends on other routers in the area known as the Designated Router (DR) and Backup Designated Router (BDR). Each non-DR and non-BDR router in the area sends Type 1 LSAs to the DR and BDR. After the DR and BDR knows about all the link states of the non-DR and non-BDR routers, then the DR updates each router within an area with route information in the form of Type 2 LSAs until ALL routers in the same area have the same network information. ABRs relay Type 3 LSAs from the DR of one area to the DR of another area. This is very important for the DR of area 0 because the ABRs connected to area 0 sends Type 3 LSAs to the DR of area 0, then the DR of area 0 shares information about other networks of other areas by sending Type 2 LSAs from the area 0 DR to the ABRs connected to area 0. Now the ABRs now knows routes to the networks of other areas. So I don't think it is accurate to say "link state doesnt depend [on other routers]!"

Source:

OSPF Design Guide - Cisco Systems -http://www.cisco.com/en/US/tech/tk365/technologies_white_paper09186a0080094e9e.shtml

NightShade1 wrote:

Look OSPF is link state inside the ARea but for interarea routeing it use vector distance
Why you dont see this in some books ? well i dont ****** know.... at least you woudlnt see that in CCNA because they just focus in OSPF in Area 0....

you got it now?
Fallow your own advice and go and read a book about it....

PD: You havent told me what means AFAIK

In an arragement of Area 1 -> Area 0 <- Area 2...

a host in area 1 forwards a packet destined for a host in area 2 to it's default gateway router.
The default gateway router which is an internal router for area 1 forwards the packet to the Area Border Router (ABR) between area 1 and area 0.
The ABR between area 1 and area 0 forwards the packet to the ABR between area 0 and area 2.
The ABR between area 0 and area 2 forwards the packet to the default gateway router in area 2 for the destination host.
The default gateway router in area 2 forwards the packet to the destination host.

Based on that scenario, since all routers are within a single OSPF Autonomous System (AS), the following routers used the SPF algorithm to create their routing tables based on the Type 2 LSAs they received from the DR of their area:

The default gateway router in area 1.
The ABR between area 1 and area 0.
The ABR between area 0 and area 2.
The default gateway router in area 2.

So in summary (a little OSPF humor there), "OSPF is a link-state routing protocol."

Cucumber

Guys, yes, OSPF is a link state routing protocol. But you must look at the big picture and try to understand how OSPF behaves when several areas are used. And then try to understand the question "Why all areas must get connected to the backbone area?".

Forget about SPF, LSA type1, LSA type2 and designated routers for a moment, move on, you already know that, that is CCNA stuff, get to the next level.

Focus on how ABRs and other LSA types work(LSA Type 3). and then you will be able to understand the question at hand (and the answer that has already been given).

Otherwise, if you want to repeat over and over OSPF is a link state protocol and never behaves as a distance vector under any circumstance, then answer the question, "why all areas must get connected to area 0?"

NightShade1

Lol tech even if RFC 1583 says that it does have a vector distance behavior.... for intra area routing.

In an arragement of Area 1 -> Area 0 <- Area 2...

1. a host in area 1 forwards a packet destined for a host in area 2 to it's default gateway router.
2. The default gateway router which is an internal router for area 1 forwards the packet to the Area Border Router (ABR) between area 1 and area 0.
3. The ABR between area 1 and area 0 forwards the packet to the ABR between area 0 and area 2.
4. The ABR between area 0 and area 2 forwards the packet to the default gateway router in area 2 for the destination host.
5. The default gateway router in area 2 forwards the packet to the destination host.

Based on that scenario, since all routers are within a single OSPF Autonomous System (AS), the following routers used the SPF algorithm to create their routing tables based on the Type 2 LSAs they received from the DR of their area:

1. The default gateway router in area 1.
2. The ABR between area 1 and area 0.
3. The ABR between area 0 and area 2.
4. The default gateway router in area 2. "

I dunno what you trying to get telling me this... explain, i know how they forward package alredy... O_o

rossonieri#1

"Lol tech even if RFC 1583 says that it does have a vector distance behavior.... for intra area routing"

You are asking about inter-area routing.

"Okay lets start hehe
Firts
area 1 --> area 0 <-- area 1 <--area 2 <--area 3

This diagram is not valid... all areas must connect to area 0..... "

if my diagram is not valid - probably my network will not connect to you right now

but - keep learning - we are all..

cheers

NightShade1

Welll those areas are still connected to area 0 via virtual links... gg....(which i said it when i told you that the diagram you put in there wasnt valid i said also it could be valid if there were a virtual links, guess you arent paying attention to what i type)

Remenber all areas must connect to the backbone; if no direct connection is possible, a virtual link may be established

Yeah we all learn new things here dont we? :P

rossonieri#1

nice discussion:)

post again if you find something new in OSPF.

cheers...

NightShade1

Well yes it is a nice discussion at least i learn a lot from this kind of discussion....
I had in other forums discussion like this one and learn a lot at the end.