Routing Protocols

Routing problem is solved on the basis of analysis of the routing tables, placed in all the routers and end nodes. How, then, is the formation of these tables? By what means is provided by the adequacy of the information contained therein is constantly changing structure of the network? The main work of creating the routing tables is performed automatically, but also the ability to manually adjust or add to the table, too, are usually provided.
To automatically build routing tables, routers exchange information about the topology of a network component in accordance with the special service protocol. Minutes of this type are called routing protocols (or routing protocols). Routing protocols (eg, RIP, OSPF, NLSP) should be distinguished from the actual network protocols (eg, IP, IPX). Both serve as the network layer model OSI - are involved in package delivery to the recipient via heterogeneous composite network. But while the first collect and transmit over the network purely proprietary information, the latter intended to carry user data, as do the data link layer. Routing protocols use network protocols as a vehicle. When exchanging routing information protocol packets are placed in the routing of data packets the network layer or even the transport layer, so in terms of investment packages routing protocols should be formally assigned to a higher level than the network.
The fact that the routers to decide on the promotion of a package addressed to the address table, you can see their similarities with bridges and switches. However the nature of their use of address tables are quite different. Instead MAC - addresses in the routing tables will be numbers of networks that are connected to the InterNetwork. Another difference between the routing tables of the address tables of bridges is a way to create them. While the bridge builds a table, passively watching the information passing through its personnel being sent to end nodes of the network to each other routers on its own initiative, communicate special service packages, informing neighbors about their well-known networks in the internetwork, routers and links of these networks with routers . Usually taken into account not only the topology of the bonds, but their capacity and status. This allows routers to adapt more quickly to changes in network configuration, as well as the right to transmit packets in networks with arbitrary topology, allowing the existence of closed loops.
With the help of routing protocols, routers make up a map network connections or less detail. On the basis of this information for each network number is decided about which next router must forward packets sent to the network to route turned out to be rational. The results of these decisions are recorded in the routing table. If you change your network configuration, some entries in the table are invalidated. In such cases, the packets sent on false trails, may get stuck and lost. On how quickly the routing protocol aligns the contents of the table to the real state of the network depends on the quality of the entire network.
Routing protocols can be built based on different algorithms, different ways of constructing routing tables, ways of selecting the best route and the other features of their work.
In all the above examples, the choice of a rational path is defined only the next (closest) router, rather than the entire sequence of routers from start to end node. In accordance with this approach, routing is performed on a distributed scheme - each router is responsible for only a single step route, and the final route is composed as a result of all the routers through which the packet. Such algorithms are called one-step route.
There is also the opposite, a multi-step approach - source routing (Source Routing). In accordance with the source node sets in the network send a full package trip to follow through all the intermediate routers. When using multi-stage routing is not necessary to construct and analyze a routing table. This speeds up the passage of the package on the network, relieving routers, but the large burden placed on the leaf nodes. This scheme is used in computer networks today are much less likely than single-step scheme of distributed routing. However, the new version of IP, along with the classical one-stop routing will be enabled and source routing.
One-step algorithms, depending on the method of forming the routing tables are divided into three classes:
- Algorithms for a fixed (or static) routing;
- A simple routing algorithms;
- Algorithms for adaptive (or dynamic) routing.
In the fixed routing algorithms, all routing table entries are static. The network administrator decides on which routers must send packets with specific addresses, or manually (eg, using a utility route on Unix or Windows NT) enters the corresponding entries in the routing table. The table is usually created at boot time, she later used without modification as long as its content will not be edited manually. Such corrections may be needed, for example, if the network denies any router and its functions are assigned to a different router. Distinguish odnomarshrutnye tables in which each recipient is given one way, and mnogomarshrutnye table defining a number of alternative paths for each recipient. In mnogomarshrutnyh tables should be set to choose one of the routes. Most often, one way is key, and the rest - redundant. It is clear that the fixed routing algorithm, with its hand-forming routing tables only acceptable in small networks with simple topology. However, this algorithm can be used effectively and to work on major highways networks, as well as the very backbone can have a simple structure with obvious best ways of following packets on a subnet attached to the backbone.
In a simple routing algorithms routing table either does not use or being built without the participation of routing protocols. There are three types of simple routing:
- Random routing, when he arrived packet is sent into the first random direction, but the source;
- Avalanche routing, the packet is sent broadcast to all possible directions, but the source (similar to processing frames bridges with unknown address);
- Routing from previous experience, when the route selection is carried out on the table but the table is based on the principle of the bridge by the analysis of address fields packets appearing at the input ports.
The most common algorithms are adaptive (or dynamic) routing. These algorithms provide automatic updating of routing tables after network reconfiguration. Protocols that are based on adaptive algorithms that allow all routers to collect information about the topology of links in the network, quickly mastering the all configuration changes relationships. In the routing tables for adaptive routing usually contains information on the time interval during which the route will remain valid. This time is called the lifetime of the route (Time To Live, TTL).
Adaptive algorithms are usually distributed in nature, which is reflected in the fact that the network does not have any dedicated routers, which would collect and summarize the topological information: This work is distributed among all routers.