ohassairi Thu, 11/01/2007 - 01:49
User Badges:
  • Silver, 250 points or more

all routers in ospf belong to the same administratif group (AS)for example one company with many branches: all routers are administered by one IT team.

in BGP we can have many authoritive teams, for example diffrent ISP.

between ISPs we use BGP.

intra ISP we can use ospf

mirzaakberali Thu, 11/01/2007 - 01:53
User Badges:

Thanks for the reply.

What makes BGP to work in larger networks efficiently than OSPF interms routing .

paul.matthews Thu, 11/01/2007 - 02:33
User Badges:
  • Silver, 250 points or more

As I mentioned, OSPF is link state. L-S does not scale to the size the internet needs. Even running link state for inter-ISP links it would be a HUGE link state database, and SPF calculations may take a while.

BGP is basically a tweaked distance vector protocol, and even that puts significant demand particularly on the memory of routers because of the number of routes in the routing table. Its neat enhancement to standard DV protocols is its loop avoidance mechanism - it tags routes with the path it has taken, so when a router sees a route that has already been through its own network, it knows to ignore it.

BGP has been designed for the job it does. as an example there are no regular/scheduled updates of routing info - it uses keepalives to maintain neighbour state, and trusts neighbour state to maintain info - you told me abut the route three weeks ago, and have stayed there on line for all that time - I do not need you to tell me again unless it changes. Most IGPs need the RP information to be refreshed periodically.

paul.matthews Thu, 11/01/2007 - 02:17
User Badges:
  • Silver, 250 points or more

They are totally different routing protocols.

OSPF is an IGP - i.e. it is used within one administrative domain. It is a link-state protocol, which can lead to more effective routing decisions. it scales if properly designed and implemented up to enterprise level.

BGP is an EGP - it is designed for routing between administrative domains. It is designed to scale in that use to planetary levels. It offers stability and scalability as two major protocol features, with control (again when properly designed and implemented) based on policy.

The two are not really interchangeable. OSPF does not offer much control within an area, and permits only one backbone - if you wth to join three administrative domains, they all need to talk to each other via area zero.

while BGP *can* work as an IGP, it does not scale well in that role - it needs too much manual influencing to work. It is normally supported by a separate IGP.

TBH comparing OSPF and BGP is a little like comparing an apple and a tomato - they are both fruit, but beyond that you would rarely use one in place of the other.


This Discussion