BGP route selection process : Step #3

[arumalpra]

I am having hard time understanding BGP route selection process stem number 3.

It sated that "Prefer the path that was locally originated via a network or aggregate BGP subcommand or through redistribution from an IGP"

Side note:

Ultimate result of BGP route selection process is

1. Give the selected prefix to router's route selection process which consider the administrative distance befor installing it in the routing table.

2. Pass it to BGP neighbor.

Say the network which is locally sourced and recieved from BGP speaker is 10.1.1.0,

few points are,

1. To use network / aggregate, this route should already be in the routing table.

2. When router recives 10.1.1.0 from BGP speaking router, it doesn't alter the routing table. It is already there withe higher administrative distance. 

(EBGP learned routes not considered as Other AS routes are not expected to learned by IGP also)

3. Locally sourced route is used in the BGP table. Therefor it is passed to bgp neighbor.

But when it is passed to other BGP router, It doesn't have any meaning whether it is originated locally or not. It only send the prefix and recieving router choose sending routes IP as the next hop.

I don't understand any significance on this step in BGP decision process as none of points in above "side note" are used in this step.

Could somebody help me to understand this or give realy world application for this?

[Mohamed Sobair]

Hello,

By Default, a BGP Speaker advertises Networks throught the following methods:

1- Once the BGP Process configured and the Neighborship is established, it advertises all Networks it recieved in its BGP Tabel, it Acts as a transient Router for All Networks if there is No Filtering in Place.

2- Through the Network or Redistribute Commands, and those Networks has to be in the routing table.

3- Through the Aggregation, Aggreagte doesnt not need to the Routes to be in the routing table, However, It requires the Networks to be in the BGP table.

So, the answer to your question, the 3rd Option in the BGP Selection Process is Valid because BGP has the Capability to recognize the Source of the Network, whether its Learned VIa  IGP or Redistribution.

Networks Learned through the BGP  Network Command are considered IGP and prefered Over the Other.

Networks Learned Through BGP Redistribution are considered (Incomplete) and has the Least preference.

Regards,

Mohamed

[arumalpra]

Actually I was refering BGP best path selcetion algorithm. starts with weitht, local preference... etc

[milan.kulik]

Hi,

"But when it is passed to other BGP router, It doesn't have any meaning whether it is originated locally or not. It only send the prefix and recieving router choose sending routes IP as the next hop."

Correct.

But don't forget the BGP Path selection algorithm is applied always on a particular router.

And from the first router point of view it has a sense to choose the prefix which was originated to BGP by the router itself - either by the network command, aggregating or IGP redistribution.

Such a prefix should should be more "accurate" than the same prefix received for any BGP neighbor, generally.

BR,

Milan

[arumalpra]

But in other words,

Router recieved prefix to the ospf process.

Same route was recieved by BGP process.

Administrative distance comes to play and 'route selection process of the router' choose OSPF path and populate "routing table".

What is the purpose of taking OSPF path over BGP  by "BGP selecton process" which populate the "BGP table" ?

[cadet alain]

Hi,

the bgp path algorithm is only for choosing the best route advertised by BGP and advertise this best route to other BGP peers and install it in the routing table if it has the best AD compared to other sources advertising it.

Regards.

Alain

Don't forget to rate helpful posts.

[milan.kulik]

Hi Alain,

I'd rephrase slightly: "The BGP best path algorithm is only for choosing the best route within the BGP table."

So if a route was received via OSPF and being chosen the best one by the 'route selection process of the router' (AD playing role), the router puts the OSPF route to the RIB.

And then it redistributes it to the BGP table (if OSFP redistribution into BGP is configured).

The BGP best path algorithm then chooses the best BGP route for that prefix (taking into its consideretion in step 3. this prefix was redistributed to BGP from IGP) and the best BGP route is advertised to the BGP neighbors.

So it's even possible (in special configurations usually) the best BGP route is different than the OSPF one (and the router does not use the BGP route for forwarding packets in fact) and still it is advertised to the BGP neighbors.

You might see an R (RIB-failed) tag for such BGP prefixes within show ip bgp command output.

See

http://www.cisco.com/en/US/customer/docs/ios/12_2s/feature/guide/fs_sbair.html

for some more details.

HTH,

Milan