Connecting to Other Autonomous Systems

Border Gateway Protocol is the only routing protocol in operation today that uses Transmission Control Protocol (TCP) for the reliable movement of its update information. TCP is the connection-oriented protocol defined at the Transport layer of the OSI reference model that provides reliable delivery of data. Routers that are configured with the BGP routing protocol are commonly called BGP speakers . Any two routers that have generated a TCP connection for sharing BGP routing information are called peers, or neighbors. These BGP peers exchange different types of messages to open and maintain the connection. BGP speakers exchange network reachability information in the form of the full paths (BGP AS numbers ) that a route should take to get to a destination network. This information is vital for building a loop-free topology where routing policies can be used to manage the routing behavior.

BGP can route between multiple autonomous systems (or domains), as well as exchange routing and reachability information with other BGP systems. BGP is often referred to as a path -vector protocol because it transports a sequence of AS numbers in its packets for loop- avoidance and designates a path to a remote network. An AS utilizes BGP to find out about other routes' reachability by gathering information from other ASs.

The rules for maintaining and exchanging information are called policies and each AS can have specific policies for the different points to external networks. BGP provides for making decisions based on rules enforced at the AS level through policy-based routing. This mechanism controls how the data is transported through the AS and is based on BGP attributes covered throughout this chapter.

BGP peers exchange their full routing tables at the outset. After the initial exchange, peers send incremental updates whenever there is a change to the routing table. BGP also stores a version number of the BGP table, which should be identical for all its peers. Whenever BGP updates the table because of a routing information change, the version number is also changed. Keepalive messages are also used to make certain that the connection is alive between the BGP peers. Notification messages are sent to respond to network errors or special situations.

BGP has four different types of messages that are exchanged between BGP routers: open, keepalive, update, and notification. Table 8.1 displays the four message types that BGP-4 uses to guarantee that updates are transmitted and neighbors remain active.

Table 8.1. BGP-4 Message Types

BGP Message	Description
Open	This message is the first message sent after the TCP connection is established. It is used to establish BGP peer relationships and contains the following fields: version, AS number of the sender, Holdtime, BGP Identifier (Router ID), and optional parameters.
Keepalive	This message confirms that connections are active or established and are exchanged between peers so that holdtimes do not expire. If holdtime intervals are configured to zero, periodic keepalives won't be exchanged.
Update	This message is generated if a change occurs and has information on a single path only. Update messages can include a list of IP address prefixes for withdrawn routes, path attributes, and lists of IP address prefixes that are reachable by this single path. Multiple paths require multiple update messages.
Notification	These messages are sent to alert BGP peers about detected errors. Notifications include an error code, subcode, and other pertinent data.

Each enterprise network or ISP on the Internet is identified by an autonomous system number (ASN) that provides a hierarchy for managing route information. The Internet Assigned Numbers Authority (IANA) is the consortium that doles out ASNs to ISPs and other organizations that use BGP (or another EGP) to connect to the Internet. The ASN is a 16-bit number (1 “65535). As with IP addresses, a range of ASNs is reserved for private use. These reside at the top of the address space from 64512 through 65535. The guidelines for ASNs are defined in RFC 1930.

Over 20 RFCs currently apply to Border Gateway Protocol version 4. For example, RFCs 1771 through 1774 define BGP-4 definition, application, and analysis. For a complete list of RFCs, navigate to www.cis.ohio-state.edu/htbin/rfc/rfc-index.html.

There are two modes of BGP: Interior Border Gateway Protocol (IBGP) and Exterior Border Gateway Protocol (EBGP). IBGP is used to exchange BGP information between BGP routers within the same AS. IGRP and OSPF are both IGPs that can be used within a BGP AS. Implementing an internal routing protocol within the AS removes the requirement that all BGP neighbors within the same AS be directly connected to one another. EBGP allows for the sharing of routing information between BGP neighbors across AS boundaries. EBGP demands that there be a direct link between neighbors that reside in different ASs, as shown in Figure 8.1.

Figure 8.1. IGPs and EGPs operating together in an internetwork design.

Path Vector

Border Gateway Protocol is often categorized as an advanced distance vector routing protocol or a "path vector" protocol. The term path vector refers to the obligatory AS-path attribute that is located in BGP update messages. This AS-path attribute tracks the incrementing sequence of autonomous systems that the packet passes through on its way to the destination network. Only EBGP speakers modify the AS-path attribute. They do this by including the ID number of the local AS to any update messages that it sends to remote ASs. The AS-path attribute, with several other attributes, is discussed later in this chapter.

Here is a list of the most important features of BGP-4:

• BGP routers exchange network reachability data, termed path vectors (or attributes), that carry a list of the full path of BGP AS numbers that a route takes to reach a destination network.

• BGP offers support for VLSM and CIDR aggregation (summarization).

• BGP utilizes TCP port 179 as its transport (Layer 4) mechanism for reliable data exchange and loop-avoidance.

• BGP exchanges the full routing table only at the outset of the BGP session.

• BGP sessions are managed and maintained with keepalive messages. Changes are propagated with update messages.

• BGP uses its own BGP table, which is completely separate from the IGP's routing table, and any network entry must first exist in the BGP table.

• BGP utilizes a rather intricate set of metrics known as attributes, which include information such as the Origin and the Next -hop address.

There is a difference between a regular autonomous system, such as Company ABC AS 65534, and one configured with a BGP transit policy, such as the ISP AS 64512. These distinctions will become much clearer over the course of this chapter and Chapter 9, "Scaling BGP Networks." However, Table 8.2 provides a good overview of the different BGP AS types.

Table 8.2. BGP AS Types

AS Type	Description
Stub AS	A stub AS is connected to only one other AS, such as a single- homed stub AS with a single connection to an ISP. These AS types often serve as enterprise BGP border ASs with a single entry and exit point linking it to an ISP's system. For that reason, stub ASs typically run an interior gateway protocol such as EIGRP or OSPF for routing internal traffic.
Transit AS	Transit ASs connect to more than one AS and operate as agents between multiple autonomous systems. For example, an ISP's AS will serve as a transit AS. The primary function of an ISP's AS is to deliver information to other ASs that belong to independent enterprise network systems. Typically, a transit AS will be configured in a fully meshed topology running only BGP as the routing protocol because no traffic is headed for intra-AS networks.
Multihomed AS	A multihomed AS is connected to multiple ASs; however, it does not operate as a transit or intermediate AS. These ASs typically link an AS to two or more ISPs to ensure that communications with Internet destinations are not interrupted if one ISP experiences a failure.

You should strongly consider implementing BGP when

• The inbound and outbound traffic flow of your network must be highly controlled.

• You need to allow traffic to pass through one AS to get to another AS.

• You are connecting multiple ISPs, network access points (NAPs), or autonomous systems.

• You are implementing a multihoming solution in an enterprise environment to connect to more than one ISP.

You should not use BGP when

• Your router does not have enough memory or processing power to handle the BGP update messages and large routing tables.

• You have only a single connection to another autonomous system or the Internet.

• You have limited knowledge of or no need for implementing routing policy or route selection and filtering techniques.

• You have low bandwidth channels between the autonomous systems.

Static routes or an IGP can be used if your organization cannot implement BGP for one of the preceding reasons.

The method for configuring static routes has already been addressed in this book. You should know the ip route command completely. For the BSCI exam, you need to know that static routing should be used instead of BGP if you meet one of the criteria in the preceding bulleted list for not using BGP. For review, the syntax for the ip route command is as follows :

  Router(config)# ip route prefix mask { address  interface} [distance]  

As you recall, when there are multiple routes to a destination network, it is the administrative distance value that determines what routes are ultimately injected into the routing table. The AD value of a static route is set to 1 if the route is designated with the next-hop address parameter. The AD value of a status route designated by the interface parameter is 0. A "floating" static route is often implemented to override a dynamically learned route. It can be used to generate a path of last resort that can be used when no dynamic route data exists. A floating static route is a sort of "path of last resort" that kicks in when no dynamic routing information is available. You can establish a floating static route by using an AD value that is higher than the default AD of the routing protocol being used. It is statically configured to be explicitly overridden by dynamically learned information.

A router running BGP maintains information concerning neighbor routers in the BGP routing table. In fact, a BGP router actually maintains two separate tables. It has the BGP routing table and the IP routing table. The BGP routing table is filled with routes obtained by the BGP protocol and the IP routing table is proliferated by routes that are obtained by IGPs running on the router. Even though these two tables are managed independently, your BGP router can be configured to exchange information between the tables. If you want to display detailed information about BGP neighbor routers, you can use the following EXEC-mode command:

 RouterA(config)# show ip bgp neighbors. 

Listing 8.1 shows an example of the show ip bgp neighbors command on a Cisco router running BGP.

Listing 8.1 A Sample show ip bgp neighbors Command

 Router#show ip bgp neighbors BGP neighbor is 172.16.11.1,  remote AS 1, external link  Index 3, Offset 0, Mask 0x8   BGP version 4, remote router ID 172.16.15.1   BGP state = Established, table version = 1, up for 01:07:28   Last read 00:00:28, hold time is 180, keepalive interval is 60 seconds   Minimum time between advertisement runs is 30 seconds   Received 70 messages, 0 notifications, 0 in queue   Sent 70 messages, 0 notifications, 0 in queue   Prefix advertised 0, suppressed 0, withdrawn 0   Default weight 10   Connections established 1; dropped 0   Last reset never   0 accepted prefixes consume 0 bytes   0 history paths consume 0 bytes Connection state is ESTAB, I/O status: 1, unread input bytes: 0 Local host: 172.16.11.2, Local port: 11000 Foreign host: 172.16.11.1, Foreign port: 179 Enqueued packets for retransmit: 0, input: 0  mis-ordered: 0 (0 bytes) Event Timers (current time is 0x18AE8E0): Timer          Starts    Wakeups            Next Retrans            71          0             0x0 TimeWait            0          0             0x0 AckHold            70         45             0x0 SendWnd             0          0             0x0 KeepAlive           0          0             0x0 GiveUp              0          0             0x0 PmtuAger            0          0             0x0 DeadWait            0          0             0x0 iss: 1359248693  snduna: 1359250034  sndnxt: 1359250034     sndwnd:  15044 irs: 1979028943  rcvnxt: 1979030284  rcvwnd:      15044  delrcvwnd:   1340 SRTT: 300 ms, RTTO: 607 ms, RTV: 3 ms, KRTT: 0 ms minRTT: 20 ms, maxRTT: 300 ms, ACK hold: 200 ms Flags: higher precedence, nagle Datagrams (max data segment is 1460 bytes): Rcvd: 97 (out of order: 0), with data: 70, total data bytes: 1340 Sent: 117 (retransmit: 0), with data: 70, total data bytes: 1340 BGP neighbor is 172.16.40.2,  remote AS 2, external link  Index 2, Offset 0, Mask 0x4   BGP version 4, remote router ID 0.0.0.0   BGP state = Idle, table version = 0   Last read 00:00:11, hold time is 180, keepalive interval is 60 seconds   Minimum time between advertisement runs is 30 seconds   Received 0 messages, 0 notifications, 0 in queue   Sent 0 messages, 0 notifications, 0 in queue   Prefix advertised 0, suppressed 0, withdrawn 0   Connections established 0; dropped 0   Last reset never   0 accepted prefixes consume 0 bytes   0 history paths consume 0 bytes   External BGP neighbor not directly connected.   No active TCP connection RouterB# 

BGP information is transported inside of TCP segments with the default TCP port designator of 179. BGP routers that have generated a BGP TCP connection-oriented communication are peers or neighbors, either internal or external. IBGP peers are routers running BGP and exchanging information within a single AS to be eventually passed on to other ASs.

BGP routers communicating between disparate ASs are directly connected and run EBGP. Before it exchanges routing table or BGP table information with a peer in an external AS, BGP makes certain that networks within the AS are reachable. This is done by a combination of internal BGP peering among routers within the AS and then redistributing BGP routing information to IGPs running within the AS, such as IGRP, IS-IS, RIP, or OSPF, as shown in Figure 8.2.

Figure 8.2. Internal and external BGP neighbors.