Why use BGP when you have an underlying protocol for connectivity ?

[myxxymoxx]

People these days seem to want to use BGP. Now BGP needs to have connectivity to their peers. So whats the point when you already have a connection from source to destination/vice versa using whatever interior protocols?

[MHM Cisco World]

the reachability between the BGP peers is use underlaying protocol which is usually ISP routing protocol, 

NOW

ISP routing protocol no need to be apart from all costumer routing protocol, so how we exchange the prefix from one costumer to other without making ISP part of this exchange ? solution is BGP

[balaji.bandi]

In most cases, IGP used to reach an end-to-end Layer 3 reachability (without any static Routing) - not a good idea for BGP to run here.

Again depends on the use case, BGP you have wider control on Traffic Engineering compare to IGP.

[myxxymoxx]

I agree here, however if you have connectivity (which BGP needs) what's the point in using it. Cannot attach my packet tracer here as it's not supported. Just trying to get my head around it. OK so we have said OSPF for connectivity, why put another protocol on top of it?

Gateway of last resort is not set

1.0.0.0/32 is subnetted, 1 subnets

B 1.1.1.1/32 [20/0] via 51.0.0.2, 00:00:00

2.0.0.0/8 is variably subnetted, 2 subnets, 2 masks

B 2.2.2.0/24 [20/0] via 51.0.0.2, 00:00:00

O IA 2.2.2.1/32 [110/6] via 71.0.0.1, 04:11:16, GigabitEthernet0/0/0

[110/6] via 41.0.0.2, 04:11:16, GigabitEthernet0/0/1

4.0.0.0/32 is subnetted, 1 subnets

B 4.4.4.4/32 [20/0] via 41.0.0.2, 00:00:00

5.0.0.0/8 is variably subnetted, 2 subnets, 2 masks

O 5.0.0.0/30 [110/2] via 41.0.0.2, 04:11:26, GigabitEthernet0/0/1

O 5.5.5.5/32 [110/3] via 41.0.0.2, 04:11:26, GigabitEthernet0/0/1

6.0.0.0/32 is subnetted, 1 subnets

O 6.6.6.6/32 [110/2] via 71.0.0.1, 04:11:36, GigabitEthernet0/0/0

7.0.0.0/32 is subnetted, 1 subnets

C 7.7.7.7/32 is directly connected, Loopback7

8.0.0.0/32 is subnetted, 1 subnets

O IA 8.8.8.8/32 [110/8] via 71.0.0.1, 04:11:16, GigabitEthernet0/0/0

[110/8] via 41.0.0.2, 04:11:16, GigabitEthernet0/0/1

9.0.0.0/8 is variably subnetted, 2 subnets, 2 masks

B 9.9.9.0/24 [20/0] via 51.0.0.2, 00:00:00

[balaji.bandi]

Just to get the clarity we are discussing Underlay and Overlay right ?

in the normal network, you do not need both. ( as i mentioned depends on use case, and scalability requirement need to use right one)

[MHM Cisco World]

EIGRP over OSPF need new IP head and so it can deploy with L3 tunnel like GRE

BUT 
BGP over OSPF no need new IP because BGP use TCP and hence no need any L3 tunnel.

[Giuseppe Larosa]

Hello @myxxymoxx ,

the answer depends on the network scenario.

In a conventional Service Provider network the IGP  ( or the IGP domains) are used to provide infrastructure connectivity that is to publish the loopback addresses of devices.

The loopback addresses are then used to build MP BGP sessions that represent customer services.

The first advantage in this context is the scalability you don't need to know the customer routes in the IGP even if they are part of the global routing table ( if using MPLS the customer destined packets travel on the Label Switched Path with destination = BGP next-hop = loopback of remote edge device, devices on the path do not route them just perform MPLS label swap ).

The second great advantage provided by BGP is the flexibility and easy of extension: MP BGP by using address families can advertise over common IPV4 messages MP NLRI that can represent entities that are not IPv4 like IPv6, VPNv4, VPNv6, different forms of L2 VPN ( VPLS with BGP for autodiscovery and LDP for pseudowires, VPLS with BGP for autodiscovery and BGP to build pseudowires), more modern services like EVPN , VXVLANs in EVPN over MPLS.

These advertisements can include additional MPLS labels to be used internally as service label.

There are other scenarios where this approach is used that in data center where combination of eBGP and  MP BGP are used to build a fabric. Again EVPN with VXVLAN can be used.

In a small enterprise network you can use an IGP.

Several great enterprise networks have adopted the service provider model and the use of MPLS and so they use MP BGP at least to easily create L3 VPNs.

Hope to help

Giuseppe