The basic idea behind OSPF

johnifanx98

Just wonder why OSPF introduces a database? Is it just for convenience of designing such a SPF algorithm?

networker050184

The OSPF database holds all the routes. The SPF algorithm is then run against that database to choose the best paths. These best paths are then placed into the routing table and used for forwarding.

Not sure if that answers your question exactly, but hope it helps.

billscott92787

Great explanation above.

johnifanx98

networker050184 wrote: »

The OSPF database holds all the routes. The SPF algorithm is then run against that database to choose the best paths. These best paths are then placed into the routing table and used for forwarding.

Not sure if that answers your question exactly, but hope it helps.

Hmmm... before I threw this post, I was not even clear what is my question. Well, thanks to your reply, I'm a bit clearer now!

The question should be,

RIP/IGRP is single area, which is imposed some limitation inherently. OSPF overcomes this weakness by introducing multiple-areas, which is good. Then, what is the necessarity of introducing a database? Why not keep to routing table like IGRP/EIGRP? Is routing table contradictory with mutiple-area?

cisco_trooper

Alright guys, I normally don't post such lame stuff but I need a pneumonic for this and it just so happens I suck at such a task.

I need a pneumonic for remembering all of the conditions that can cause a CEF Punt.

Any contributions are appreciated.

cisco_trooper

Wow, I can't believe I just posted that here...my apologies for having my head where it doesn't belong.

EdTheLad

johnifanx98 wrote: »

RIP/IGRP is single area, which is imposed some limitation inherently. OSPF overcomes this weakness by introducing multiple-areas, which is good. Then, what is the necessarity of introducing a database? Why not keep to routing table like IGRP/EIGRP? Is routing table contradictory with mutiple-area?

RIP/IGRP are distance vector protocols that learn about routes using a technique called routing by rumor.Routing by rumor means the protocol relies on third party information from its directly connected neighbors.This info can take a while to propagate through the network so convergence time increases, also due to not knowing the complete topology loops can occur which need to be controlled using holddown timers which again increase convergence time.

OSPF is link state which means every router in an area receives routing info from every other router and creates a database containing the path/distance to every other router and network within the area.This database is constant throughout an area so every route in an area has exactly the same database.Each router runs the SPF calculation on the database putting itself as root.The result of the SPF calc will be a Shortest Path Tree which contains the route/cost to every network in the area, these route are placed in the routing table.

If the area is huge the database will be huge, the SPF calculation will take along time and more RAM will be required on each router to hold the database.To decrease the calc time, areas can be divided into smaller area's and summary routes can be exchanged beween Area Border Routers.Since an ABR will be in more than one area it will have more than one database.

So OSPF knows the complete topology of the network which means loops within an area cannot occur and holddown timers are not necessary.If a link fails OSPF can run a quick recomputation on its database and insert a new route into the RIB without having to wait for route poisoning to occur.

You need to read more on OSPF as its very clearly described in the Cisco Press guides.