BGP (routing between Autonomous Systems) Protocol

　　

1. Introduction

BGP is a routing protocol between autonomous systems. The Network accessibility information exchanged by BGP provides sufficient information to detect routing loops and make decisions based on performance precedence and policy constraints. In particular, BGP exchanges the network accessibility information of all as paths and executes routing policies according to the configuration information.

With the development and growth of the Internet in recent years, it has to face some serious scale problems, including:

-Class B network address space is exhausted. One of the main reasons for this problem is the lack of medium-size networks suitable for medium-sized organizations; a class C network can have a maximum of 254 host addresses, which is too small, the Class B network allows a maximum of 65534 addresses, but it is too large to be fully used.

-The increasing number of Route tables in Internet Routers makes the current software (and people) unable to be effectively managed.

-32-bit IP address space depletion.

Obviously, the first two problems and the last one may become urgent in the next one or two years or three years. Classless Inter-Domain Routing (CIDR) attempts to solve these problems and designs a mechanism to reduce the growth of Route tables and new IP Address allocation requirements. It does not solve the third longer-term problem. Instead, it tries to delay the recent problem so that the Internet can still operate effectively and start a long-term solution.

The BGP-4 expands the BGP-3, and supports the aggregation of routing information and the reduction of Routing Based on classless Inter-Domain Routing System (CIDR. This memorandum discusses the application of BGP-4 in Internet.

All the discussions in this document are based on the following assumptions: the Internet is a set of autonomous systems that are randomly connected. That is to say, the Internet can be modeled into a general network diagram. The nodes on the graph are as, and the edges are connections between each pair of.

The classic definition of an autonomous system is that a group of routers use the internal gateway protocol and unified measurement to route data packets under unified management in the, the external gateway protocol routes data packets to other. The classic definition is still evolving, and some as uses a variety of internal gateway protocols and measurements within it. Here, we emphasize the meaning of autonomous systems in this document. Even if it uses multiple IGP and metrics, its management is different from that of other as systems, and its internal routes are consistent. When a route passes through it, it is regarded as a node on the graph. Each as is managed by a management organization, at least externally it represents the routing information of the system.

2. BGP topology model

When we talk about a connection between two as, it means two things:

Physical connection: A Shared data link subnet exists between two as instances. On this subnet, each as instance must have at least one VBR. Therefore, each as Border Gateway Router can forward data packets to other as Border Gateway routers without routing between AS and.

BGP connection: There is a BGP session process between the BGP spokespersons of each as. Through the session communication route, the declared as reaches a target network.

In this document, we impose additional restrictions on the BGP spokespersons that form BGP connections: they must share their own data link subnets directly. Therefore, BGP sessions between adjacent as instances do not require routes between as instances or as instances. Cases beyond the scope of this article may be inconsistent with this limit.

Therefore, in each connection, each as has more than one BGP speaker and Border Gateway Router, which are distributed on the Internet of the shared data link. Note that the BGP spokesman is not necessarily a Border Gateway Router, and vice versa. The path stated by the BGP spokesman connecting to the same as can be used by other as Border Gateway routers on the same shared subnet, that is, the non-direct neighbors are allowed.

The traffic in an as is either from the as or finally from the as (that is, the source IP address or destination IP address of the IP packet is in the ). The traffic that meets the preceding description is called "local traffic"; otherwise, it is called "transitional traffic ". The main purpose of BGP is to control the transitional traffic.

As can be divided into the following categories based on how an as handles transitional traffic:

End as: connect to only one other. Naturally, the end as only ships local traffic.

Multi-host as: connects to more than one other as, but does not transport the transitional traffic.

Transition as: connects to more than one other as to transport local and transitional traffic.

A complete as path provides an effective and simple way to avoid routing loops and eliminate the "count to infinity" problem of the companion Distance Vector Algorithm. Therefore, BGP does not limit the connection topology between as instances.

3. BGP in the Internet

3.1 Topology Considerations

The Internet topology can be considered as any connection between the transition as, multi-host as, and end. To minimize the impact on the current Internet structure, BGP does not have to be used for the end and multi-host. These as can run other protocols (such as EGP) to exchange network accessibility information with the transition. The transition as using BGP will mark the information to indicate that it learns methods other than BGP. BGP does not necessarily run on the end or multi-host as, which does not negatively affect the routing quality between as data packets originating from or finally from the end as or multi-host.

However, we recommend using BGP on the end and multi-host. In these cases, BGP provides better bandwidth and performance than other currently used protocols (such as EGP. In addition, this reduces the use of default routes and provides a better choice for multi-host as routes.

3.2 overall BGP features

At the overall level, BGP is used to transmit route information between multiple Autonomous Systems. Its information flow is shown as follows:

　

<ccid_code style="margin: 0px; padding: 0px;">　 　 +-----------+　+------------+BGP　|　 BGP　 |　 BGP　 |　 BGP |　 BGP　 -------------------+ +--------------------------+ 　 |　 IGP　 | 　|　 IGP |　 　 +------------+ +-----------+<-AS A-->　<--AS B->

This chart shows that only BGP is used to transmit information between as, while BGP and IGP can both transmit information within. It is a major issue to ensure the compatibility of routing information between BGP and IGP in.

3.3 BGP neighbor relationship

The Internet is regarded as a set of freely connected. The router directly connected through BGP is the spokesman for BGP. The BGP spokesman can be in the same as or different. Each as BGP spokesman communicates with each other and exchanges network accessibility information according to the policies established by each. For a BGP spokesman, if it communicates with other BGP spokespersons and the BGP spokesman is in a different as, the other BGP spokesman is called an external peer. If it is in the same, it is called an internal peer.

There can be many BGP spokespersons in an as and are considered necessary. Generally, if one as has multiple connections with other as, multiple BGP spokespersons are required. All BGP spokespersons represent the same as and maintain a unified image externally. This requires that they maintain consistent routing information. These routers can communicate with each other through BGP or other methods. Policy constraints of all BGP spokespersons within the same as must be consistent. Some technologies, such as using labeled IGP, can be used to detect possible contradictions.

For external peers, the peers belong to different as, but share the same data link subnet. This common subnet is used to transport BGP information between peers. If BGP uses an as that uses an intervention, the as path information is invalid. The autonomous system number must be used in BGP to indicate the autonomous system where the BGP spokesman is located.