Analysis and Comparison of common routing protocols

Dynamic Routing varies with the network operation. The router automatically calculates the optimal data transmission path based on the functions provided by the routing protocol, and obtains the dynamic route table.

Based on the Routing algorithm, dynamic Routing protocols can be divided into Distance Vector Routing Protocol (Distance Vector Routing Protocol) and Link State Routing Protocol (Link State Routing Protocol ). The distance vector routing protocol is based on the Bellman-Ford algorithm, mainly including RIP and IGRP (IGRP is the private protocol of Cisco). The Link State routing protocol is based on the Dijkstra algorithm, which is very famous in graph theory, that is, the Shortest Path First, SPF algorithm, such as OSPF. In Distance Vector routing protocol, routers transmit some or all route tables to their adjacent routers;

In the link status routing protocol, the router transmits the link status information to all routers in the same region.

Based on the position of the router in the Autonomous System (AS), the routing Protocol can be divided into the Internal Gateway Protocol (IGP) and the External Gateway Protocol (EGP, is also called the Inter-Domain Routing Protocol ).

There are two types of Inter-Domain Routing Protocols: the external Gateway Protocol (EGP) and the Border Gateway Protocol (BGP ). EGP is designed for a simple tree topology. It has obvious disadvantages when dealing with routing loops and setting routing policies. It has been replaced by BGP.

VPN is a private protocol of Cisco. It is a hybrid protocol. It not only has the characteristics of distance vector routing protocol, but also inherits the advantages of Link State routing protocol. Different routing protocols have their own characteristics and are suitable for different types of networks. The following sections describe them respectively.

Static Routing

The static route table is created by the network administrator before you start to select a route, and can only be changed by the network administrator. Therefore, it is only applicable to environments with simple network transmission statuses. Static Routing has the following features:

Static Routing does not require route switching, Saving network bandwidth, CPU utilization, and vro memory.

Static Routing provides higher security. In a network using a static route, all routers connected to the network must set their corresponding routes on the adjacent router. Therefore, network security is improved to some extent.

In some cases, you must use static routes, such as DDR and the network environment using NAT technology.

Static Routing has the following Disadvantages:

Managers must understand the network topology and correctly configure routes.

Poor network scalability. To add a network to the network, the Administrator must add a route to all routers.

Configuration is cumbersome, especially when it is necessary to communicate across several routers, the routing configuration is more complex.

Dynamic Routing

Dynamic Routing Protocols include distance vector routing protocol and Link State Routing Protocol. These two protocols have their own characteristics and are described as follows.

1. Distance Vector (DV) Protocol

Distance vectors refer to the number of hops or vectors used to determine the distance from one device to another. The speed of each hop link is not considered.

The Distance Vector Routing Protocol does not use normal neighbor relationships. Two methods are used to obtain the topology change and route Timeout:

When the router cannot directly receive route updates from the connected router;

When a router receives an update from its neighbor, it notifies the router of a change in its network topology somewhere.

In a small network (with less than 100 routers or less routing updates and computing environments), the vector routing protocol runs quite well. When a small network is extended to a large network, the convergence speed of the new route is extremely slow for the algorithm to calculate, and the network is in a transitional state during its calculation, loops are likely to occur and cause temporary congestion. Furthermore, when the underlying network link technology is diverse and bandwidth is different, distance vector algorithms turn a blind eye to this.

This feature of the Distance Vector routing protocol not only results in network convergence delay, but also consumes bandwidth. As the route table increases, more CPU resources and memory are consumed.

2. Link Status (LS) Routing Protocol

The link status routing protocol does not limit the number of hops. The "graphics theory" algorithm or the Shortest Path priority algorithm is used.

The link status Routing Protocol has a shorter convergence time, supports VLSM (Variable Length subnet mask), and CIDR.

The link status routing protocol maintains a normal neighbor relationship between directly connected routes. This allows faster route convergence. The link status Routing Protocol creates a peering relationship by switching the Hello packet (also called the link status information) during a session, which accelerates the convergence of the route.

Unlike the distance vector routing protocol, the entire route table is sent during update. The link status routing protocol only broadcasts the updated or changed network topology, which makes the update information smaller and saves bandwidth and CPU usage. In addition, if the network does not change, the update package is sent only within a specific period of time (usually 30 minutes to 2 h ).

3. Comparison between link status routing protocol and Distance Vector Routing Protocol

Analysis of common dynamic routing protocols

4.1 RIP

RIP (Routing Information Protocol) is the first open standard used by router manufacturers. It is the most extensive routing protocol and can be obtained on all IP routing platforms. When RIP is used, a Cisco router can be connected to a router of another manufacturer. RIP has two versions: r1_1 and r1_2, which are based on the classic distance vector routing algorithm and have a maximum of 15 hops.

R12001 is a Classful Routing protocol. Because mask information is not included in the route, all devices on the network must use the same subnet mask and VLSM is not supported. R12002 can send subnet mask information. It is a non-family Routing (Classless Routing) protocol and supports VLSM.

RIP uses UDP data packets to update route information. The router updates the route information once every 30 s. If no response is received from the neighboring router within S, the route to the router is considered unavailable and the router cannot arrive. If you do not receive a response from the vro after S, delete the route information about the vro from the route table.

RIP has the following features:

Routers of different manufacturers can be interconnected through RIP;

Simple configuration;

Suitable for small networks (less than 15 hops );

R1_1 does not support VLSM;

Wan bandwidth required;

CPU and memory resources are required.

The RIP algorithm is simple, but the convergence speed is slow when many paths exist. When route information is broadcast, many bandwidth resources are occupied, it is suitable for small networks with relatively simple network topology and extremely low data link failure rate. In large networks, RIP is generally not used.

IGRP

The Interior Gateway Routing Protocol (IGRP) was developed by Cisco in 1980s. It is a dynamic, long-span Routing Protocol that supports up to 255 hops, measure (vector) is used to determine the optimal route to a network. The optimal route is calculated by latency, bandwidth, reliability, and load. It has a high span in the same autonomous system, suitable for complex networks. Cisco IOS allows the router administrator to set weights for the network bandwidth, latency, reliability, and load of IGRP to influence the calculation of the metric.

Like RIP, IGRP uses UDP to send route table entries. Each vro updates the route information every 90s. If no response from a vro is received within S, the vro cannot be reached. If no response is received within S, the IGRP process deletes the route from the route table.

Compared with RIP, IGRP has a longer convergence time, but the bandwidth required for transmitting route information is reduced. In addition, there are no blank bytes in the IGRP grouping format, which improves the message efficiency of IGRP. However, IGRP is proprietary to Cisco and is only applicable to Cisco products.

China site

With the expansion of the network scale and the growth of user requirements, the original IGRP has been unable to cope with the problem, so Cisco has developed an enhanced IGRP, that is, iver6. VPN uses the same routing algorithm as IGRP, But it integrates the strengths of the link status routing protocol and distance vector routing protocol, and adds the spread update algorithm (DUAL ).

VPN has the following features:

Fast Convergence. Fast Convergence is achieved by backing up routes in the routing table using the propagation update algorithm, that is, the minimum overhead and the minimum overhead (also called feasible successor) to reach the destination network) routes are saved in the routing table. When the route with the minimum overhead is unavailable, the route is quickly switched to the next Least overhead route to achieve fast convergence.

Reduces bandwidth consumption. Unlike RIP and IGRP, VPN gateway exchange route information once every time. When the Route status of a destination network changes or the measurement of a route changes, therefore, the bandwidth required for route update is much lower than that required by RIP and OSPF. This method is called triggered ).

Increase the network size. For RIP, the maximum network size is 15 hops, while the maximum network size is 255 hops ).

Reduce the CPU usage of the router. Route updates are only sent to the neighboring routers that need to know the status change. Because incremental updates are used, fewer CPUs are used than IGRP.

Supports a variable-length subnet mask.

IGRP and VPN can be automatically transplanted. IGRP routes can be automatically re-distributed to the network in the network, and the routes can be automatically re-distributed to the network in the network. If you want to, you can also disable route redistribution.

VPN gateway supports three routing protocols (IP, IPX, and AppleTalk ).

Supports load balancing with non-equivalent paths.

Because EIGIP is a dedicated protocol developed by Cisco, when a Cisco device is connected to a device from another manufacturer, you cannot use

4.4 OSPF

Open Shortest Path First (OSPF) is an internal gateway routing protocol developed for IP networks. It is developed and recommended by IETF. OSPF consists of three subprotocols: Hello protocol, exchange protocol, and diffusion protocol. The Hello protocol is responsible for checking whether the link is available, and completing the specified vro and the specified vro for backup. The Exchange Protocol completes the designation of the "master" and "slave" vro and exchanges the information of their respective routing databases; the diffusion Protocol completes the synchronization maintenance of the routing database in each vro.

The OSPF protocol has the following advantages:

OSPF can represent the entire network in its own link status database, which greatly reduces the convergence time and supports interconnection of large heterogeneous networks, it provides a way for a heterogeneous network to exchange network information through the same protocol and is not prone to incorrect routing information. · OSPF supports multiple paths to the same destination.

OSPF uses routing labels to distinguish different external routes.

OSPF supports route verification. Only routers that pass route verification can exchange route information. In addition, different verification methods can be defined for different regions to improve network security.

OSPF supports load balancing on multiple links at the same cost.

OSPF is a non-family routing protocol. The route information is not limited by the number of hops, which reduces the subnet separation problem caused by hierarchical routing.

OSPF supports query of VLSM and non-family routes, facilitating effective management of network addresses.

OSPF uses AREA to layer the network, reducing the CPU processing time and memory requirements for the Protocol.

4.5 BGP

BGP is used to connect to the Internet. BGPv4 is an external routing protocol. It can be considered as an advanced distance vector routing protocol.

In a BGP Network, a network can be divided into multiple Autonomous Systems. The autonomous systems use eBGP broadcast routes, while the autonomous systems use iBGP to broadcast routes in their own networks.

The Internet is composed of multiple interconnected commercial networks. Each enterprise network or ISP must define an autonomous system number (ASN ). These autonomous system Numbers are allocated by IANA (Internet Assigned Numbers Authority. A total of 65535 Autonomous System numbers are available, of which 65512 ~ 65535 is reserved for private use. When the route information is shared, this number can also be maintained in a layer-by-layer manner.

BGP uses reliable session management. Port 179 in TCP is used to trigger Update and Keepalive information to its neighbors to spread and Update the BGP Route table.

In BGP networks, autonomous systems include: 1. Stub

There is only one portal and one egress network.

2. Transfer AS (Transit)

When data is transferred from one AS to another AS, it must go through Transit.

If the enterprise network has multiple AS, you can set Transit AS in the enterprise network.

The biggest difference between IGP and BGP is that the total number of additional information passed between devices running the protocol is different. The route update package used by IGP is smaller than the route update package used by BGP (therefore, BGP carries more routing attributes ). BGP can be attached with many attributes on a given route.

When two routers running BGP start to communicate to exchange dynamic routing information, TCP port 179 is used. They rely on connection-oriented communication (session ). BGP must rely on connection-oriented TCP sessions to provide the connection status. Because BGP cannot use Keepalive information (but Keepalive information is stored in common headers to allow the router to check whether the session is Active ).

The standard Keepalive is the information sent from one vro to another on the circuit without using TCP sessions. The router uses these signals to check whether the circuit is correct or not. In some cases, BGP is required:

When you need to send traffic from one AS to another;

When the data flow from the outbound network must be manually maintained;

When you connect two or more ISPs, NAP (Network Access Point), and switching point.

BGP cannot be used in the following three cases:

If your router does not support the large route table required by BGP;

Use the default route when there is only one connection to the Internet;

When your network does not have enough bandwidth to transmit the required data (including the BGP Route table ).