How to select a Routing Protocol

When the routing protocol is enabled, the network has the powerful function of automatically updating the route table. However, some major internal gateway protocols (InteriorGatewayProtocol, IGP), such as RIP/RIP2, OSPF, or IGRP/VPN, have certain protocols.

The Internal Gateway Protocol is applicable to scenarios where only one Administrator is responsible for network operations and operations. Otherwise, a configuration error may cause network performance degradation or unstable network operation. For networks shared by many administrators, such as the Internet, consider using the EGP protocol (InteriorGatewayProtocol, external gateway protocol), such as BGP4.

If there is only one vro in the network, no routing protocol is required. It is necessary to share information only when there are multiple vrouters in the network. However, if you only have a small network, you can use static routes to manually update the route table.

Routing Information Protocol

The RIP (RoutingInformationProtocol) protocol is based on a program called "routed". The program runs on a BSDI Unix system and is standardized in RFC1058 in 1988. In version 2 described in RFC1388, support for VLSM (VariableLengthSubnetMasks) is added, but it does not make up for the major defects of the Protocol. For example, in a network with multiple paths to the same destination, it takes a lot of time for RIP to determine a selectable path. In a network without multiple paths, the RIP Protocol has been widely used.

The RIP Protocol is listed as the distance vector protocol, which means that distance is used to determine the optimal path, as measured by the number of Route hops. The router sends broadcast information to each other every 30 seconds. Each vro that receives the broadcast information adds a hop count. If the broadcast information is received by multiple routers, the path with the lowest hop count on the router is the selected path. If the preferred path does not work properly, the path with a higher hop count will be used as a backup.

For the RIP Protocol (and other routing protocols), the router on the network must go through the process of deciding the alternative path when a path is unavailable. This process is called Convergence ). The main problem is that the RIP Protocol spends a lot of time on convergence. Before the RIP Protocol realizes that the path cannot be reached, it is set to wait until it has missed 6 updates for a total of 180 seconds. Then, before using the new path to update the route table, it waits for the next information of another feasible path to arrive. This means that it takes at least three minutes before the backup path is used, which is a long time for most applications to time out.

Another basic problem of the RIP Protocol is that it ignores the connection speed problem when selecting a path. For example, if a path composed of all fast Ethernet connections is a hop farther than a path containing a 10 Mbps Ethernet connection, the path with a low 10 Mbps Ethernet connection will be selected as the best path.

The original version of the RIP Protocol cannot apply VLSM, so it cannot split the address space to apply limited IP addresses to the maximum efficiency. RIP2 implements this function by introducing subnet shielding and using each channel by broadcast information.

The routing protocol should also prevent data packets from entering the loop or routing selection loop. This is because redundant connections affect the network. The RIP Protocol assumes that if more than 15 routes are jumped from one terminal of the network to another, loops are involved. Therefore, when a path reaches 16 hops, it is considered to be inaccessible. Obviously, this limits the use of the RIP Protocol only on the network.

The biggest problem with RIP involves a large network with redundant paths. If there is no redundant path in the network, the RIP Protocol works well. It is an Internet standard implemented by almost every vendor that supports path selection. The RIP Protocol is applicable to most server operating systems and is easy to configure and fix obstacles. Other routing protocols should be considered for large networks or networks with redundant paths.

OSPF2

OSPF2 is an Internet standard similar to the RIP Protocol and can make up for the shortcomings of the RIP Protocol. It was first standardized in RFC1247 in 1991; the latest version was in RFC2328. However, unlike the RIP Protocol, OSPF is a set of link status routing protocols.
The changes in route selection are based on the physical connection status and speed of the vro in the network, and the changes are immediately broadcast to each vro in the network.

When an OSPF router is activated for the first time, it uses the OSPF "hello protocol" to discover the adjacent nodes connected to it, and then uses LSA (link status broadcast information) and exchange link status information with these routers. Each vro creates a database consisting of each interface, corresponding adjacent node, and interface speed. The LSA received by each vro from the neighboring vro continues to be transmitted to the neighboring vro until each vro in the network receives the LSA from all other vrouters.

The link status database is different from the route table. Based on the information in the database, Each router calculates a path to each destination of the network and creates a route topology tree with it as the root, it contains the Shortest Path priority tree (SPF tree) that forms the basis of the route table ). LSA is switched every 30 minutes, unless the network topology changes. For example, if the interface changes, information is immediately broadcast through the network; if there are redundant paths, convergence will recalculate the SPF tree. The time required to calculate the SPF tree depends on the size of the network. Because of these calculations, the router needs to use more CPU resources to run OSPF.

One way to make up for the OSPF protocol to occupy CPU and memory resources is to divide the network into independent hierarchical domains, called areas ). Each vro only exchanges LSA with other vrouters in their own regions. Area0 is used as the trunk area, and all areas must be connected adjacent to Area0. The boundary between two regions is defined on the ABR (Regional Border Router, AreaBorderRouter. There must be at least one interface for the ABR, Area0, and another non-trunk region. The OSPF network with optimal design includes the backbone network that is adjacent to each region through VLSM. This makes it possible to describe multiple networks in an entry in the route table.

Although the OSPF protocol is a powerful alternative to the RIP Protocol, it requires more router resources for execution. If the network is running the RIP Protocol and there is no problem, you can continue to use it. However, if you want to use redundant links based on standard protocols in the network, OSPF is a better choice.

Enhance the routing protocol of the Internal Gateway

In Cisco products, the EnhancedInteriorGatewayRontingProtocol has some advantages. The most important thing is that it can rapidly broadcast changes in the link status. However, the biggest disadvantage of the Protocol is that there is no standardization.

Like the OSPF protocol, the OSPF routers look for their adjacent routers and exchange "hello" packets. The OSPF protocol transmits "hello" packets every five seconds. If the router fails three times, the adjacent router is considered as in the down state, and the alternative path is used.

When the link status of the local router changes, it recalculates the topology table based on the new information. At this time, the OSPF protocol immediately broadcasts changes in the link status to each vro in the network. The OSPF protocol only involves the vro directly affected by these changes. This increases the utilization efficiency of bandwidth and CPU resources. At the same time, because the VPN protocol uses less than 50% of the bandwidth, it has a great advantage in the low-bandwidth WAN link. Another advantage of the Protocol is that it supports the Novell/IPX and AppleTalk environments. If the network is running the IGRP protocol, it is much easier to convert to the OSPF protocol.

Article entry: csh responsible editor: csh