Help me in BGP concepts

[Amr Ali]

Dear All,

I hope every one in a good condition , kindly i need an assistance as i am a beginner in CCNP routing so i need to understand many of things , i have uploaded a topology made by GNS3 which had done by me but i need to understand why R1 can't ping R5

Thanks for Help

[John Blakley]

By default, intra-area and inter-area routes are redistributed when doing redistribution, so in order to get everything (external routes), you specify external type-1 and external -type 2. "Match internal" matches inter and intra area routes which is the default.

Redistribute bgp 5500 subnets tells ospf to redistribute bgp routes from the as 5500. The "subnets" keyword tells OSPF to pull all routes even if not classful. By default, it will only redistribute classful subnets.

HTH,
John

*** Please rate all useful posts ***

[Amr Ali]

thanks john for help

This is the rest of my post as i need to understand these notes" plus  i need to understand why i must provide BGP  with another routing  protocol like OSPF or EIGRP to complete routing  & i need to know  when i must use cmd # Network at BGP & cmd #  Neighbour & the difference bet. them & why we have used cmd #  Network at R5 only and didn't use it at any other router "

Thanks for your time & Help

Regards - amr

[John Blakley · Accepted solution]

"i need to understand why i must provide BGP  with another routing  protocol like OSPF or EIGRP to complete routing"

Well, you don't. You can have just BGP in your environment and it will work fine, or you can run an IGP like ospf or eigrp all over the place and it will work fine as well. The primary difference is that BGP was designed to be able to hold hundreds of thousands of routes if you have enough memory to hold a full routing table. An igp on the other hand was designed to route internal subnets, so you wouldn't, in general, have hundreds of thousands of routes. An IGP can't handle the same amount of routes that bgp can.

"i need to know  when i must use cmd # Network at BGP & cmd #  Neighbour"

Network command under bgp allows for you to advertise the specific network. The network needs to be in the routing table in order to advertise though, and bgp needs an exact match. 10.10.10.10/32 is not the same as 10.10.10.0/24 under bgp. BGP would need "network 10.10.10.10 mask 255.255.255.255" in order to advertise the 10.10.10.10/32 address. The neighbor command tells bgp who to connect to to share those addresses with by peering up with them. BGP works on a unicast address and has a direct connection to another peer. By default bgp has a ttl of 1, which means that it has to be a directly connected peer. You can disable that check though if you're peering with someone on the other side of a next hop or a loopback address.

"why we have used cmd #  Network at R5 only and didn't use it at any other router "

R5 was running BGP, but wasn't advertising anything. Technically, I'd have to lab this up again when I get back home to test it, but you probably didn't need it after all and here's why. R5 was connected directly to R4 10.1.45.0, so R4 knew how to get to R5 and vice versa. The problem is that R1 thought to get to 10.1.45.0, it needed to go to 4.4.4.4. 4.4.4.4 went to R2 or R3, but R2 or R3 didn't know about 10.1.45.0 and dropped the packet. So, in reality, you could have kept your peering with R1 --> R4 --> R5 and only advertise the 10.1.45.0 in ospf on R4. You would still have to redistribute into BGP so R5 would know how to get to your other subnets unless you wanted to advertise a default route to R5 from R4.

HTH,
John

*** Please rate all useful posts ***

[Amr Ali]

Dear John ,

Thank you so much for your amazing knowledge , last thing i need to compare between cmd # network in OSPF & in BGP , in addition i need to understand what does it mean by " RIB Failure " & what is it's cure as i have added an extra Router R6 to the topology & it can ping normally R1 but when i used cmd # show ip bgp   at R1 i found all networks had a RIB Failure so when i used cmd # show ip bgp rib failure   the reason was Higher Admin Distance although R1 can ping any other networks and working normally

Cheers - amr

[John Blakley]

Amr,

A rib failure for the reason that you're seeing with "higher admin distance" means that there router has a better route than what BGP can provide. In your case on R4, anything learned from R5 will have an AD of 20 due to ebgp. Anything on R1 learned from R4 via bgp will have an AD of 200 and anything on R1 learned from ospf will have an AD of 110. So if you have a route that OSPF matches, for example 10.1.45.0, you would have a rib failure. (If I remember right, I think you can see this on R4 for the 10.45.1.0/24 network that R5 advertises.) R4 would show a rib failure because 10.45.1.0 is a connected network for it being an AD cost of 0. Does that make sense? There's nothing wrong with rib failures and  you'll see them in production networks. If you were learning an ospf route of 1.1.1.0/24 and a bgp route of 1.1.1.0/24, the bgp route would have a rib failure due to high admin distance, but if the ospf route were to go away for some reason, the bgp route would be entered into the routing table.

HTH,
John

*** Please rate all useful posts ***