Ues of Hello on OSPF virtual-link

wave

What is the point of the Hello parameter in the OSPF area X virtual-link command? The Hellos are suppressed so what's the point in specifying value for Hello?

martell1000

as far as i understood this hello packets are being suppressed as soon as the state goes to FULL on a virtual link amd are just used to establish the adjacency.

note that any changes you make to the OSPF authentication do not take effect until you clear the OSPF process in this case.

wave

My thoughts too.

Rather than starting another thread...Let's say you have the following topology:

Router A (area 0) --- Router B (area 1) --- Router C (area 2)

You create a virtual-link from C to A. Router C advertises a summary of 172.16.0.0/16 into area 0 and area 1. So Router C is injecting the summary route of 172.16.0.0/16 into area 0 and so is Router B.

The summary route from router C, over the virtual-link, will have the lowest cost because Router B is technically invisible so its interfaces aren't included when calculating the cost right?

martell1000

(area 0) Router A (area 1) --- (area 1) Router B (area 1) --- (area 1) Router C (area 2)

this is the way it kinda looks with interfaces, installing a virtual link will be done between ROUTER A (router id) and ROUTER C (router id)

the link will create a tunnel which makes router c act like it was directly connected to area0. router b just acts a a part of the tunnel and doesn't care much about the virtual link.

wave

martell1000 wrote: »

(area 0) Router A (area 1) --- (area 1) Router B (area 1) --- (area 1) Router C (area 2)

this is the way it kinda looks with interfaces, installing a virtual link will be done between ROUTER A (router id) and ROUTER C (router id)

the link will create a tunnel which makes router c act like it was directly connected to area0. router b just acts a a part of the tunnel and doesn't care much about the virtual link.

Yes yes, I understand how a virtual-link works. But what I was wanting to confirm is that the cost of the summary route from area 2 would be lower than the cost of the same route being passed on by Router B.

In the ROUTE FLG page 277 it states that the network attached to Router C (they call it R3) will be injected into both area 1 and area 0 in a type 3 LSA. That means you have two separate LSAs about the same network coming from Router C over the virtual link AND Router B. Both are sending the LSA into area 0. So then Router A in area 0 is obviously going to see two routes to the same network:

1 over the virtual link
2 through Router B

My assumtion is that the path via the virtual link will have the lowest cost. I plan to lab this but wanted to throw it out for confirmation now.

martell1000

oh i see what you mean.

but as you can see the 172.19.0.0 network in area 2 is being advertised in area 0 ONLY by router 3.3.3.3
the networks in area 1 are being advertised by 1.1.1.1 AND 3.3.3.3 because the are both ABRs which have an interface in that area

on page 278 they show how the metric differences in such a case look (from the point of R3 this time):

advertisments from router 1.1.1.1 come with a metric of 75
and as 3.3.3.3 has an interface in area 1 also the metric here is 10

so the whole point of this is that r3 becomes an ABR with interfaces in 3 different areas and doesnt have to route everything over the virtual link but can install links to area1 into his routing table using his interface connected to area1

at least this is how i understood it

wave

martell1000 wrote: »

oh i see what you mean.

but as you can see the 172.19.0.0 network in area 2 is being advertised in area 0 ONLY by router 3.3.3.3

You're right. I checked the output again on p 277 and the route to 172.19.0.0 coming from 3.3.3.3 only - in area 0. I was thinking that R2 could re-advertise a route it received in a type 3 LSA from R3 but it can't. An ABR advertises (packs and sends type 3 LSA) for its networks only.

It all makes sense because all areas are supposed to connect back to area 0, also why Cisco only want us to use a virtual-link if we absolutely have to.

Thanks martell1000

martell1000

your welcome. i have been studying ospf for the route exam for the last weeks and looking into this topic again certainly was a good thing

wave

martell1000 wrote: »

your welcome. i have been studying ospf for the route exam for the last weeks and looking into this topic again certainly was a good thing

I know what you mean! I've finished the OSPF chapter and labbed a bit as I went. Now I'm working my way through OSPF labs in the Cisco Networking Academy lab book. Last night I suddenly wondered where certain bits 'n pieces of EIGRP knowledge had gone, I had to flip back through my notes and the FLG to refresh my memory. I guess this is only to be expected when you're pounding away at any given protocol.

My plan is to read each chapter, labbing a bit as I go but doing most of my lab work when I've finished reading the chapter.

martell1000

haha

went the same road in the last day. i feel like an ospf boss atm because i was pushing this topic for weeks. but recognized eigrp and bgp and IPV6 are also there. so i am jumping into EVERYTHING right now...

route is sceduled for next weeks wednesday - and i am scared like ****

what are your plans for the route exam? always feels good to talk to someone going through the same **** atm

wave

Good luck!!

I started studying around mid January and I think it will take me about 3-4 months in total to study and pass. Probably closer to four. When did you start studying?

I get a good two and a half hours of study done Monday to Friday and the odd lunch hour. On the weekends I can usually get a solid 4-5 hours in each day.

I used the official Boson Ex Sim max practice exams for ICND 1 & 2 and I plan to do the same for all CCNP exams. Do you use them? I found them to be incredibly helpful. Never really been into flashcards.

martell1000

thank you

i started in last august. but then life happened and i didnt do really much till christmas. never used boson stuff, for ccnp i got cbt nuggets, the flg book, gns3, the ine videos and several websites with gns3 labs on it. think i might be ready for the exam but you`ll never know till you sit there

wave

When an ABR detects a link failure I've read it floods an LSA for the failed network with a cost of infinity. I assume that on point-to-point links in single area OSPF or in a broadcast single area the same would remain true? i.e. The router/network type LSA would be flooded to all routers in the area with a cost of infinity?

From watching the debugs an LSU is definitely sent. I'm just wanting to confirm that the determining "link down" factor is the cost set to infinity...essentially, does it do poison-reverse

EDIT: It appears originating router will set the LSA to maxage (60min) so it will be immediately deleted. (I guess flooded to other routers first, then deleted.)