An Insight on Routing

Routing is the process by which a group determines the network scope of an end-to-end path from the source to the destination. Routing works on the third layer of the OSI Reference Model-the network layer's packet forwarding device. Routers implement network interconnection by forwarding packets. Although routers can support a variety of protocols (such as TCP/IP, ipx/spx, AppleTalk, etc.), most routers in China run TCP/IP protocol. Routers typically connect two or more logical ports identified by an IP subnet or Point-to-Point protocol, with at least one physical port. The router determines the output port and the next hop address based on the network layer address in the packet and the routing table maintained within the router, and overrides the link layer packet header to implement forwarding packets. Routers reflect the current network topology by dynamically maintaining the routing table, and maintain the routing table by exchanging routing and link information from other routers on the network.

Brief introduction

Routing is the activity of transmitting information from the source address to the destination site through an interconnected network. Routing occurs in the third layer, the network layer, in the OSI Network Reference Model. Route-Guide packet forwarding, passing through some intermediate nodes, to their final destinations. As a hardware, it is called a router. Routing typically directs packet forwarding based on the routing table-a table that is stored to the best path for each destination. Therefore, in order to efficiently transfer packets, it is important to establish a routing table stored in the router's memory. The difference between routing and bridging lies in the similarity of the node distance between the routing and the assumed address. This allows a record in the routing table to represent the path to a group of addresses. Therefore, in a large network, routing is better than bridge, and routing has become the most important way to find a path on the Internet. Smaller networks can usually manually set the routing table, but a larger network with complex topologies can often change, and it is impractical to manually establish a routing table. However, most public switched telephone networks (PSTN) still use a predefined routing table to use a prepared path when the path of a direct connection is disconnected; see Public switched telephone network routing. "Dynamic Routing" attempts to automatically set up a routing table to solve this problem by using the information that is carried by the routing protocol, and allows the network to avoid network disconnection or failure almost autonomously. Dynamic routing currently dominates the entire Internet. However, it often requires experience and technology to set routing protocols, and the current network technology has not yet been developed to automatically set routing. Packet-switched networks (such as the Internet) divide the data into a number of packets with a full destination location, each of which is forwarded separately. Circuit-switched networks, such as the public Switched telephone network, also use the route to find a path that allows the next data to arrive at the correct destination only with a partial destination location.

Basic concepts

Routing is the activity of transmitting information from the source address to the destination site through an interconnected network.

Routing occurs in the third layer, the network layer, in the OSI Network Reference Model. Route-Guide packet forwarding, passing through some intermediate nodes, to their final destinations. As a hardware, it is called a router. Routing typically directs packet forwarding based on the routing table, a table that is stored to the best path for each destination.

Therefore, in order to efficiently transfer packets, it is important to establish a routing table stored in the router's memory. The difference between routing and bridging lies in the similarity of the node distance between the routing and the assumed address. This allows a record in the routing table to represent the path to a group of addresses. Therefore, in a large network, routing is better than bridge, and routing has become the most important way to find a path on the Internet.

Smaller networks can usually manually set the routing table, but a larger network with complex topologies can often change, and it is impractical to manually establish a routing table. However, most public switched telephone networks (PSTN) still use a predefined routing table to use a prepared path when the path of a direct connection is disconnected; see Public switched telephone network routing.

"Dynamic Routing" attempts to automatically set up a routing table to solve this problem by using the information that is carried by the routing protocol, and allows the network to avoid network disconnection or failure almost autonomously. Dynamic routing currently dominates the entire Internet.

However, it often requires experience and technology to set routing protocols, and the current network technology has not yet been developed to automatically set routing. Packet-switched networks (such as the Internet) divide the data into a number of packets with a full destination location, each of which is forwarded separately. Circuit-switched networks, such as the public Switched telephone network, also use the route to find a path that allows the next data to arrive at the correct destination only with a partial destination location.

Dynamic routing

If a set path is not available, the existing node must decide on another path to send the data to the destination. They usually use the following two types of routing protocols to achieve: distance vector algorithm and the wiring state algorithm.

Almost all routing algorithms can be categorized into these two algorithms.

1. Distance vector algorithm

Main entry: Distance vector routing protocol The distance vector algorithm uses the Bellman-ford algorithm. For paths between nodes on each network, the algorithm assigns a "cost" to them.

The node selects the lowest path for the total cost (the sum of all costs through the path), which is used to send the data from the node to the Node B. This algorithm is very simple. When a node first starts, it knows only its neighbor node (the node directly connected to that node) and the cost to that node. (This information, the list of destinations, the total cost of each destination, and the "next node" that must go through to a destination form the routing table, or the distance table.) Each node periodically sends information about the cost of the current knowledge to each destination to each neighbor node. Neighbor nodes examine the information and compare it to what is currently known, and if the cost to a destination is lower than currently known, the information received is added to your routing table. After a while, all nodes on the network will know the best "next node" and the lowest total cost for all destinations.

When a node is disconnected, each node that treats it as a "next node" of a path discards the routing information and creates a new routing table information. Then they tell all the neighboring nodes, and find the new path to all the destinations that can be reached.

2. Wired State algorithm

Main entry: Wired Status routing protocol In the wired state algorithm, each node has an atlas of the network (a graph). Each node transmits other node information that it can connect to all nodes on the network, and the other nodes then add the information to the map.

Each router can determine the best path from itself to the other node based on this map. The algorithm that completes this action--dijkstra algorithm--build another kind of data structure--tree. The node-generated tree treats itself as the root node, and the last tree will contain all the other nodes in the network. Initially, this tree has only the root node (the node itself).

Then, in the neighbor node of the existing node in the tree and not in the node collection in the tree, select a node with the lowest cost to join the tree until all nodes are stored in the tree.

This tree is used to establish the routing table, provide the best "next node" and so on, so that nodes can communicate with other nodes in the network.

Comparison of folded routing algorithms

In a small network, the distance vector routing protocol is simple and efficient, and requires only slight management. However, their scale is not good, and the convergence is very poor, thus promoting the development of a more complex, but much better, link state routing protocol for use in larger networks.

The distance vector routing protocol also has an infinite count problem (count-to-infinity problem, see [1]). The main advantage of the Wired State routing protocol is that it responds more quickly to line changes, such as disconnection, within a limited time. Furthermore, the packets sent by the Wired State routing protocol on the network are smaller than the packet of the distance vector routing protocol. The distance vector routing protocol must transmit the entire routing table of a node, but the packet-state routing contract only needs to transmit the neighbor information of that node. As a result, these packets are so small that they do not occupy significant network resources. The main disadvantage of wired state routing protocol is that it requires more storage space and stronger computational ability than distance vector routing protocol.

To send a contract around

Sometimes routing agreements and routable agreements are often confusing: Can be routed protocol: Any network protocol that provides sufficient network-level address information to allow packets to be transferred from one device to another without the need to know the entire path from the source to the destination. The can be wrapped contract defines the format of the package and how the Packet field is used. Packets are usually delivered from one terminal system to another.

IP is an example of a protocol that can be bypassed, and an Ethernet path that is not to be bypassed. Routing protocols: Exchanging routing information between networks, allowing routers to dynamically establish a protocol for routing tables.

Traditional IP routing is simple because it uses the next node routing method, which means that the router only needs to consider which "next node" to send the packet to, without taking into account the entire path of the destination.

Although dynamic routing can be very complex, it makes the Internet very resilient and allows the internet to grow more than eight orders of magnitude since the adoption of IP. The routing metric (routing metric) contains all the values that are used by the routing algorithm to determine which path is better than the other path. Metrics can include a lot of information, such as bandwidth, latency, through-holiday points, path costs, load, MTU, reliability, and transmission costs. The routing table stores only the best possible paths, but the connection state or the topology database may store other related information.

When routers find multiple paths that can reach the same destination from different routing protocols, they choose the best path using attributes called management distance (administrative distance). The management distance defines the degree of reliability of routing protocols.

Each routing protocol distinguishes precedence by the most reliable to least reliable arrangement, according to the management distance value.

There are many kinds of routing protocols in accordance with the relationship between routers and other autonomous systems: Ad hoc network routing protocols appear on a network that has no or little foundation.

See the Ad Hoc Routing agreement list for the proposed agreement. The internal Gate Protocol (IGPS) exchanges routing information in a single autonomous system. Common examples include the following:

IGRP (Interior Gateway Routing Protocol)

EIGRP (Enhanced Interior Gateway Routing Protocol) [1]

OSPF (Open shortest Path a)

RIP (Routing information Protocol)

Is-is (intermediate system to intermediate system) The external Gate Protocol (EGPS) exchanges routing information in different autonomous systems.

EGP includes:

EGP (External Gate Agreement, formerly used to connect the backbone of the Internet, is now no longer in use) BGP (Border Gateway Protocol: Use the current version around 1995, BGPV4).