Routing is the function of finding the closest and possible routes between all networks in a network topology. Since the main purpose of this function is to find the best route, it is necessary to first define which is the best route and which metric should be used to measure the route.
What is Routing in Computer Networks?
The routing function represents the path to a group of devices of a single route table entry for configured addresses, assuming network addresses are configured and similar addresses are in proximity within the network.
In large networks, structured addressing outperforms unstructured addressing, and routing is widely used as it is the basis of addressing on the Internet. However, bridging is still a widely used solution in local area networks.
The network metric is the number of hops required to go from one point to the destination, initially this value is 1 for all connections and increases by one as the network grows.
Another type of metric is the expression of metric in units of time, that is, the measurement of the latency of the data traffic between neighboring nodes where network values are not constant but depend on network traffic.
The network metric values mentioned above are used to determine which is the best route to a target network and can have different protocols and different measurement values.
In cases where a router has more than one path to the target network, some methods can also be used to determine which of the routes closest to the target network is best. In some cases, the router may determine that the best route is the less distant one and select the route with the best bandwidth for the best route.
In this case, depending on the protocols used in the network environment, the metric calculation system for finding the best way here may differ from each other.
The Best Way
Calculation of the best route is determined by different structures and protocols, but can usually be determined by considering the delay value and the minimum hop count between network nodes.
Also, determining the best path depends on the shortest path to the destination, ie choosing the path through the least number of nodes. Generally, the distance or cost of a network refers to the quality of the link based on the defined metric.
Virtual Circuit and Datagram Networks
The packet-switched network operates in a virtual circuit model, and the route function usually creates a path through the lifetime of this virtual circuit. In this case, the route process between networks is managed by the session.
A network operating in datagram mode does not guarantee the regular transmission of packets and therefore nodes can change the route conditions for each packet to be sent. In this way, the algorithm can discover the optimal route for a new network, and an alternative route is created when there is any change in the topology.
Classification of Algorithms
Static routes do not take into account the state of the subnet to find the best route and are manually configured for route tables between networks. Also, static route tables are not deleted from the router, unless they are changed or deleted by an administrator.
Therefore, they do not provide real-time integration support for changes in network conditions or failure conditions. The complexity of the algorithm and the time required for convergence are not important as there is no computation of the best route, but it is higher in terms of performance.
They become more adaptable to changes in the subnet, such as dynamic route changes in network traffic, increased latency, or failures in topology, and provide rapid convergence.
This type is divided into three groups, central adaptability, distributed adaptability, and isolated adaptability, depending on where network decisions are made and the source of information exchanged.
In its central adaptive structure, all nodes in the network are a central node that collects control information and data from other nodes to calculate the route table. The disadvantage of this method is that it consumes plenty of resources from the network itself.
The purpose of the distributed adaptive structure is to run the respective algorithm equally on all nodes of the subnet, and each node relies on this information and the continuous recalculation and updating of the table based on the data contained in its database.
The most widely used routing algorithms on the Internet, the distance-vector algorithm and the link-state algorithm belong to the dynamic structure.
In an isolated structure, the simplicity of the method used to adapt to the changing state of the network is handled, and the information in each node’s own and local database is used in traffic or topology changes.
Dynamic Route with Distributed Algorithms
Distributed algorithms and route operations are based on the structure of the dynamic model. Algorithms work with the latest data the network receives about the nodes, and they ensure rapid convergence by optimizing new routes. In addition, the route tables automatically adapt to changes in the network and extreme traffic loads.
These types are divided into two main groups according to this type of operation;
1. Distance Vector
The distance vector uses the Bellman-Ford algorithm and finds the lowest cost route using the indirect search method and sends the packet to the destination. It is a control packet that calculates the distance to network nodes in relation to a network node.
Each node sends the distances it knows to its neighbors through this packet, and neighbor nodes examine this information and update their tables to compare with what they already have. Examples of distance vector protocols; RIPv1 and v2, IGRP, and EIGRP.
2. Link State
The link-state network structure is based on each node’s topology of the network and the latency values associated with the connections and uses the Dijkstra algorithm. Examples of this type are OSPF and IS-IS protocols.
On the Internet, an autonomous system (AS) consists of a set of IP networks and routers that are under the control of the same entity and have a similar route policy. There are different protocol classifications depending on the relationship of a router to an autonomous system.
Ad Hoc Protocols
These protocols are used in networks with little or no infrastructure.
Interior Gateway Protocols
Interior gateway protocols are used in systems that transfer tables within a single autonomous system. For example;