What is BGP and How Does it Work?

What is an Autonomous System (AS)?

Autonomous System (AS) is a group of interconnected networks connected to internet routers that are managed by a single administrative domain, such as a business, organization, or enterprise.

It is assigned by a 2-byte number called ASN or Autonomous Number. These ASNs are exhausted now, so the Internet Engineering Task Force (IETF) came up with a 4-byte ASN.

Type of Autonomous System (AS)

There are three types of Autonomous systems:

1. Multi Home AS - AS connected to two or more Autonomous Systems

2. Stub AS - AS connected to only one Autonomous System

3. Transit AS - AS acts like a link between two or more Autonomous Systems

BGP Routing

BGP is a path vector protocol, which means it chooses the best route to a destination based on the qualities and path information of a route.

Internet BGP routers exchange these routes which they are aware of with each other across several autonomous systems.

Based on these routes, internet BGP routers build a routing table that displays the most direct route to each destination.

What is the BGP Port Number?

BGP is an application layer protocol that uses TCP at the transport layer. Its TCP port number is 179.

What is BGP neighbor or peering?

Internet routers should know each other before sharing any information.

So first these routers become BGP neighbors or peers with each other if both peers successfully pass BGP authentication(md5 method), you may also be required to perform troubleshooting and only then they start sharing routing information effectively.

Later they also exchange routing updates if there are some changes in the connected networks.

What are BGP States?

Let's learn about BGP states in brief, these are the different phases in the process of routers becoming BGP neighbors. 

1. Idle State: In this initial state, the two routers who want to become neighbors wait for the BGP process to start.

2. Connect State: In this state, TCP connection requests initiate but a 3-way handshake is not completed yet.

3. Active State: Enter into this state when BGP is unable to establish a TCP connection with the neighbor.

4. OpenConfirm State: When a BGP router receives an Open message from its neighbor and also sends Keepalive. 

5. Established State: When both BGP routers receive Keepalive messages from each other. Here they start exchanging routing information.

How AS PATH Attribute Handled by eBGP and iBGP

It’s important to remember that although if the two types of BGP are utilized in various ways, they share a common protocol and perform similarly.

The way EBGP and IBGP handle the AS PATH parameter is the major distinction between the two protocols. I cover BGP attributes in the later part of this blog article.

When a route is learned in EBGP, the AS PATH attribute is updated to include the router’s AS number. It helps in loop avoidance in eBGP.

While AS number is not assigned to the AS PATH element in iBGP.

What is Split Horizon in BGP?

Split horizon is a mechanism used in iBGP to avoid routing loops. If an iBGP router receives a route from its iBGP peer, then the iBGP router does not propagate that route to its other iBGP peer. Routes are not passed on to the other peer creating a routing issue. There are some solutions to this problem:

1. Full Mesh: All routers in a AS form iBGP peering with each other.

2. Route Reflector: All routers form iBGP peering with a Route Reflector (RR).

3. Confederation: Large AS is divided into sub-as

BGP Path Attributes

The optimum route to a destination is chosen by BGP routers via a technique known as BGP Best Path Selection.

Route attributes like the Weight, AS PATH, the origin code, the MED, etc. are used to determine which route to choose. The optimum path is determined by the router to be the one with the lowest metric.

BGP Best Path Selection Criteria

BGP routers select the best path among multiple paths available to reach to the destination networks. The BGP best path selection criteria are based on the attributes associated with each route. The criteria, in order of preference, are as follows:

1. Weight: It's a Cisco-specific attribute, a route with a higher weight is preferred. It is local to a router. 

2. Local Preference: It determines the preference of a route within AS. A BGP route with a higher preference value is preferred. The default local preference value is 500.

3. AS Path Length: An AS is prepended when a route is traversed through various BGP autonomous systems. A BGP route with a shorter AS Path length is preferred.

4. Origin Code: IGP routes are preferred over EGP, and EGP routes are preferred over incomplete routes.

● IGP routes are advertised using interior gateway routing protocols and determined by "i".

● EGP routes are the routes advertised by exterior gateway protocol (Older BGP - never used).

● Incomplete routes are the routes advertised using redistribution and determined by "?".

5. Multi-Exit Discriminator (MED): It determines the preferred exit point for traffic leaving the AS, Lower BGP MED attribute configuration value is preferred.

6. eBGP vs iBGP: eBGP routes are preferred over iBGP routes.

7. IGP metric to Next Hop: A route with a lower IGP metric to the next hop is preferred. 

8. Next Hop: A route with a reachable next hop IP address is preferred.

9. Router ID: BGP uses the Router ID of the router advertising the route. The router with the lowest Router ID is preferred.

10. Cluster ID and Originator ID: They are considered only in the case of route reflectors. A route with the lowest Originator ID is preferred and an Originator ID is used to avoid loops.

What is BGP Prefix List?

Prefix list is a route filtering mechanism used in BGP and with other protocols based on IP address prefixes. A prefix list is created by a name and with a sequence number for example

!

ip prefix-list TEST-LIST seq 10 permit 10.0.0.0/8

ip prefix-list TEST-LIST seq 20 permit 20.0.0.0/8

!

In this prefix list, there is a default implicit deny statement at the end of the prefix list. These prefix list can directly be used in BGP neighbor statement inbound or outbound to filter IP address prefixes. As an example, below is the BGP configuration.

!

router bgp 100

  neighbor 192.168.1.1 remote-as 101

  neighbor 192.168.1.1 prefix-list TEST-LIST out

!

In the above configuration, prefix-list TEST-LIST is applied outbound to allow only the IP address prefixes present in the TEST-LIST prefix list to its BGP neighbor 192.168.1.1

What is the BGP Route Map?

The prefix list can also be called from a route map for example below, there may be many other parameters such as local preference, weight, access-list, med, etc. can be called from a route map. The route map statements are also permitted and denied with a sequence number. 

!

route-map MY-ROUTE-MAP permit 10

 ip address prefix-list TEST-LIST

!

Now this route map then can be applied to the neighbor statement under BGP process. 

!

router bgp 100

  neighbor 192.168.1.1 remote-as 101

  neighbor 192.168.1.1 route-map MY-ROUTE-MAP out

!

In this configuration, route map also has an implicit deny statement in the end. The route-map is applied on neighbor 192.168.1.1 outbound, this route-map has prefix list which is permitting 10.0.0.0/8 and 20.0.0.0/8 only to its neighbor. 

Note: In above case, the result achieved by both i.e. directly applying prefix list and using route-map are same. 

What is BGP Community?

BGP community is a 32-bit attribute, represented as 16-bit pair value for example 65000:101, first 16-bit value is AS number and other 16-bit value is assigned by administrator. There are the following types of communities.

1. Transitive vs. Non-transitive:

Transitive communities are propagated to BGP neighbors, while non-transitive communities are not. The default behavior is transitive.

2. Well-Known Communities:

There are a few well-known BGP communities with predefined meanings. For example:

No-Export : Prevents the route from being advertised to external BGP peers.

No-Advertise: Prevents the route from being advertised to any BGP peer.

3. Custom Communities:

Network administrators can define and use custom communities based on their specific needs. These communities can be used to implement various policies or convey information about the route.