cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1004
Views
5
Helpful
6
Replies

cisco bgp implemetation

mlu
Level 1
Level 1

in the book of <<internet routing architecture>>, page 159, stated "With Cisco IOS, the RID is the loopback address if one is configured; otherwise, it's the highest IP address on the router".

but in the RFC 1771:

"- otherwise, select the route that was advertised by the BGP speaker whose BGP Identifier has the lowest value."

in the RFC draft "http://www.ietf.org/internet-drafts/draft-ietf-idr-bgp4-24.txt"

"g) Prefer the route received from the lowest peer address."

seem to me, cisco implemetation was a bit different from RFC or just a misprint of halabi's book?

6 Replies 6

Hello,

I don´t see a contradiction here, Halabi talks about the way a BGP router choses its router ID, while the RFC talks about the tie breaker rule within an AS. Where do you see the contradiction ?

Regards,

Georg

the tie breakers of the route selection process is for all ajacency-RIB-in tables in a BGP router.

if loopback address is not configured, then the RID should be the smallest numerical peer address. but cisco uses the highest numerical address in the router as last resort.

Juniper uses the samllest numerical ip address of a peer router.

seem to me, juniper sticks to the RFC on this part.

There's nothing that I can see in the BGP specification that deals with what a router's RID should be. Since the concept of an RID isn't BGP-specific, it wouldn't make sense for the BGP specification to address it -- how an RID is chosen by any given router is orthogonal to how the BGP decision process uses that RID.

Incidentially, Juniper's documentation says that "the first hardware interface" is used as the RID if a specific RID isn't configured and there's no loopback IP address configured: http://www.juniper.net/techpubs/software/junos/junos63/swconfig63-routing/html/routing-generic-config3.html.

i don't think i was talking about the RID issue here. maybe i should use the phrase "BGP identifier" instead of RID.

i was wondering why cisco implemetation choose the highest ip address as its RID (juniper uses lowest) if no loopback configured (no RID configured).

But there's nothing in the BGP RFC or current draft that addresses what a given router's BGP ID should be. Although a lower BGP ID wins in the route selection algorithm, the RFC/draft doesn't require that the router use its lowest IP address as its BGP ID -- it doesn't even make a recommendation in this regard.

Juniper's documentation says that the first interface detected (not necessarily the one with the lowest address) is used for the BGP ID as well as the RID: http://www.juniper.net/techpubs/software/junos/junos63/swconfig63-routing/html/bgp-config6.html

i was aware of the none-specific nature of rid issue. what i was trying to learn the differences between two implememtations and how those differences would impact the process of bgp route selections.

if my memory serves me correct, cisco ospf implemetation also choose the highest ip address as its rid; like cisco bgp implemetation, if no rid nor loopback configured.

thanks. the discussion real helps me understand more on the implemetations of bgp.

the rid configuration in juniper is optional and it could be done on the cisco routers too:

router bgp xxxxxx

bgp router-id a.b.c.d

Review Cisco Networking for a $25 gift card