Routers can be hardware or software and are used to distribute network packets between different networks in a network. The best known routers are probably the Internet routers that connect the home network to the Internet. But not only the Internet connection but also other applications are very common, for example the use for a Virtual Private Network (VPN) or for the connection of individual network parts e.g. LAN and WAN.
Routing works according to the OSI reference model on the network layer layer. Also known as Layer 3, the mediation layer regulates the communication between terminals and the network.
Part of the routing is used for addressing terminal devices. They use the IP protocol, more precisely the network address portion of the IP address. A router has at least one physical or virtual interface for connecting networks.

Many routers can also translate private and public IP addresses (NAT/PAT, Network Address Translation, Port Address Translation) and can also act as a firewall. Sometimes policy-based routing can also be used for routing, where in addition to the destination address, other aspects of the data packet such as the origin (IP address), special requirements or other parameters are used. This makes it possible to route Web packets (HTTP) in a different way than, for example, mail packets (SMTP).

How does the routing work?

The router finds the correct destination address of a data packet, e.g. by means of the network address within the IP address, always trying to find the shortest and fastest way to the destination address. In order to determine where the individual addresses are located within a network, the network device uses the routing table. A routing table can be understood as an address book for destination addresses in a network, it prevents which route is to be used via which interface is to be used to reach the target. The entries in the routing table are partly created automatically by the router, partly they have to be entered manually. Directly connected devices appear automatically in the routing table if configured correctly. On the other hand, static routes must be manually entered into the table by an administrator. Dynamic routes are routes learned using the routing protocol. The routing protocol contains growing information about the participants of the network and about the network itself. If there is no specific entry for a destination address, a default entry is used.

Do not confuse the routed protocols (e.g. IP or IPX) and the routing protocols. Routing protocols manage the routing process and are part of the router. Routed protocols are the protocols of the data packets that are transported. The following applies: Not all protocols can be transported, but only data packets from protocols such as IPv4, IPv6 or IPX/SPX can be transported. Other protocols can be routed through tunnels and special features such as data link switching, using switches and bridges. However, this capability is not self-evident for every router, only a part works as Layer 3 switches, i.e. they can perform bridge functions. In the case of larger local networks, an attempt is also made to keep the broadcast volume for the individual parts small. For DHCP and other broadcast services to work with the router, the router must have a relay agent function.

What types of routers are there?

According to its protocol abilities one can distinguish different types:

Single protocol routers are only suitable for network protocols such as IPv4. Multiprotocol routers (or multiprotocol routers) can handle multiple protocol families, e.g. IP, DECnet, IPX/SPX and others.

However, network protocols play only a minor role today apart from the IP protocol, since they can also be transmitted encapsulated via the IP protocol, e.g. as IPX or NetBIOS over TCP/IP.

There are also useful categories for distinguishing between the different types:

Hardware routers, also known as backbones, can be found on the Internet or at large companies; they are optimized for transporting or mediating different data packages. They have a very high data throughput and are designed for permanent fail-safe operation. No single central processor is loaded, but the computing power is provided decentrally by network interfaces. Several data packets can be transported in parallel, since the ports and interfaces transmit data independently of each other. Some manufacturers also refer to hardware routers as switches with routing functionality. These devices also specialize in transporting data packets, but use the MAC address instead of the IP address.

Normal PCs can also be used as routers. In this case, the routing function is performed at operating system level and using the CPU. With such software routers, high power consumption is to be expected. POSIX operating systems already have a routing function, but other operating systems can also be used for router operation via additional software. Many operating systems already have built-in extensive routing capabilities, such as Microsoft Windows, Mac OS X Server or Linux.

Depending on the orientation of the application, different devices are used. Large companies often use hardware or software routers that are optimized for robustness and longevity and can be used in control cabinets.
Pure software routers are mainly used in the not professional area, either as a dedicated router or as a non-dedicated router. If a workstation or server is used almost exclusively for routing, it is called a dedicated router, while the non-dedicated variant also performs other functions in addition to routing.

Hardware routers are used in the high-end sector, they have special high-performance buses or cross bars and are designed for continuous operation and reliability. They are clearly superior in the performance of a software variant, but are also very expensive to purchase. Simple PCs do not have the necessary hardware components, but better servers or workstations can also have redundant components and thus ensure reliability in some areas.

It is important to know that some devices known as hardware routers are composed of PC components and only a few special parts, such as the housing or special PCI slots, have been processed. However, such systems still perform software routing.

If performance is not a critical factor, both dedicated and undedicated systems can be used. Professional solutions are always inferior to both forms in terms of performance.
The reason for this is that a classic PCI bus with 32-bit bus width and a clock rate of 33 MHz is often used, which could theoretically transport 1 GBit/s in principle. Since the transfer takes place in half-duplex mode (HDX), the network packets are routed twice over the PCI bus, so that the transferred data stream is automatically reduced by half in each direction. However, Ethernet today is operated in full duplex mode (FDX), which a system with a 32-bit PCI bus and 33 MHz cannot achieve at all. If systems with a 64-bit PCI bus and 66 MHz are used, then can achieve these over the approx. 4 GBit/s. Better server systems have even faster interfaces (e.g. PCI-X 266) and several independent PCI buses. With such systems one can achieve considerably higher throughput rates, but then also has a high energy consumption, which in turn is reflected in high operating costs. If one has sufficient financial means, hardware copies with CPUs and chipsets aligned accordingly are certainly the better choice with regard to the energy consumption.

In order to achieve high routing performance, however, you do not necessarily need dedicated hardware, but you can also use a routing cluster. The cluster can be composed of several software routers (for example, on several workstations). The software routers are connected to each other via switches with a sufficient number of ports and a data throughput rate of several hundred GBit/s. The routers are connected to each other via a network of switches. However, it should be noted that the maximum data throughput rate of the entire routing cluster can only ever correspond to the maximum throughput rate of the central switch. In addition, the clusters can be designed redundantly, making them particularly high-performance, fail-safe and cost-effective.

DSL routers have a PPPoE client for dialing into the Internet with DSL. DSL routers have Network Address Translation (NAT) in IPv4 networks so that public IPv4 addresses can translate into private IPv4 addresses. DSL variants can also have a switch, WLAN and a telecommunications and VoIP system.

Border routers are mostly used by Internet service providers to connect participants’ networks to other autonomous systems. Usually the External Border Gateway Protocol (EBGP) is used.

WLAN routers are combinations of switch, wireless access point and router and have a WAN port for the connection of two or more WLAN devices. moreNets. This allows WLAN and WAN routing to take place between the networks or between LAN and WAN routing. Often a device specialized on WLAN can only be used for a small area of application.

Keine Produkte gefunden.