BGP Synchronization

Bosefus

I am not grasping the concept of BGP sync.

From what I understand, the synchronization rule states that BGP cannot advertise a route unless it learns/hears the route from an IGP.

Does this apply to internal and external routes?

For instance, if I two edge routers in my AS connected to two separate ASes outside of my AS. If one edge router hears about a network from another AS, will it not pass the route to its BGP peer in its own AS because of the synchronization rule?

I read another example which further confused me.

BGP Case Studies - Cisco Systems

According to the doc, RTB will not advertise the route to AS400 until it hears about the route from RTE.

In the example, it states that there is no redistribution running, so how will RTB here about the route from an IGP(RTE)?

Thank you for the help.

kryolla

This is from the case study

Synchronization states that, if your AS passes traffic from another AS to a third AS, BGP should not advertise a route before all the routers in your AS have learned about the route via IGP

so it is for external routes only. It also states you can put a static route on RTB as long as it is in the routing table it will advertise it out to an eBGP neighbor.

BGP relies on IGP to get to the next hop of the final destination so if IGP or the routers in the path to the next hop doesnt know about the final destination it will drop the packet and send an unreachable out. You can turn it off but it still doesnt solve the underlying issue. The next hop has to recurse to an outgoing interface or it will not be put in the routing table.

So you get a BGP route X via next hop of Y which recurses to an outgoing interface of fa0/0 and send it out to a directly connected neighbor
The neighbor which is not running BGP then looks in its routing table for the final destination and will say it is unroutable since it doesnt have route X and send an icmp unreachable back.

So why would a BGP router send the route to another AS if it is just going to black hole that traffic.

Bosefus

So you get a BGP route X via next hop of Y which recurses to an outgoing interface of fa0/0 and send it out to a directly connected neighbor
The neighbor which is not running BGP then looks in its routing table for the final destination and will say it is unroutable since it doesnt have route X and send an icmp unreachable back.

So, then in order for the next hop router to be able to route the packet you must either configure a static route to the destination or turn on redistribution, correct?

networker050184

Bosefus wrote: »

So, then in order for the next hop router to be able to route the packet you must either configure a static route to the destination or turn on redistribution, correct?

Or turn off synchronization which I believe is the default now.

kryolla

Bosefus wrote: »

So, then in order for the next hop router to be able to route the packet you must either configure a static route to the destination or turn on redistribution, correct?

yes in order to advertise it to an eBGP neighbor but

I dont think you are understanding the underlying issue. Sure you can turn off sync and it will advertise the route to an eBGP peer like what networker050184 said but you are still going to black hole traffic. ALL routers in the path of the next hop need to know about the final destination or it will drop the traffic.

Bosefus

So , my IGP would need to know paths to all external routes that come into the AS in order for BGP to advertise them out to another AS, is this correct?

If so, wouldn't the routing tables for hte IGP be huge if you had a lot of external routes?

Thanks

kryolla

if you want to be used as a transit AS then yes if not then just advertise you own routes out. Keep in mind these are routes that you received via eBGP and advertise it out to another AS via eBGP

Edit: you can set up an GRE tunnel between the 2 iBPG neighbors which will also work if you leave sync off. But more than likely for an enterprise you wont advertise other AS routes.