Network topology

The topology of a computer network referred to in the structure of the connections of several units with each other to provide a common data exchange.

The topology of a network is crucial for its reliability: Only if alternative paths between the nodes exist in failures of individual compounds the functionality is retained. There are in addition the way to work one or more substitute routes (or diversions ).

The knowledge of the topology of a network is also useful for the evaluation of its performance and the investment and for the selection of suitable hardware.

A distinction is made between physical and logical topology. The physical topology describes the structure of the network cabling; the logical topology of the data flow between the terminals.

Topologies are displayed graphically (after the graph theory) represented with nodes and edges.

In large networks, it is often a structure composed of a number of different topologies.

• 2.1 point-to -point topology 2.1.1 benefits
• 2.1.2 disadvantages

• 2.2.1 benefits
• 2.2.2 disadvantages
• 2.2.3 Examples

• 2.3.1 benefits
• 2.3.2 disadvantages
• 2.3.3 Examples

• 2.4.1 benefits
• 2.4.2 disadvantages
• 2.4.3 Examples

• 2.5.1 benefits
• 2.5.2 disadvantages
• 2.5.3 k- tree
• 2.5.4 Ring Extended Tree
• 2.5.5 hypertree
• 2.5.6 Fat Tree

• 2.6.1 benefits
• 2.6.2 disadvantages

• 2.7.1 benefits
• 2.7.2 disadvantages
• 2.7.3 Examples

• 2.8.1 Star - bus
• 2.8.2 Star Star

Characteristics

Diameter

The diameter of a topology describes the maximum straight-line distance between two nodes in hops. So that it is a direct measure for the expected maximum transfer time, that is, the larger the diameter, the greater the transfer time in the worst case.

Degree

The degree of a topology is the number of links per node. These may be the same or different for each node. Do all nodes have the same degree of a topology, the topology is regular, which is beneficial to the network. Further, the degree describes indirectly the costs you have to spend to build the topology. The higher the level, the higher the cost.

Bisektionsweite

The Bisektionsweite specifies the minimum number of links that must be traversed to share a network with N nodes in two networks, each with N / 2 nodes. It is thus a measure of the performance of a network as the nodes of a network half communicate in many algorithms with the nodes of the other half. So the lower the Bisektionsweite, the more unfavorable will be affected on the time required for the data exchange between the two halves power.

Symmetry

In a symmetric topology, the network of any viewpoint sees (nodes / links) from the same, ie there exist for nodes and / or edges so-called graph - automorphisms. In simple terms, this means that nodes and / or links in a symmetric network behave the same no matter which node or link which you look. This has a significant positive effect (simplification ) on programming, load balancing and routing, since there are no special cases to consider.

Scalability

Scalability indicates the smallest Netzinkrement ( number of nodes and links ) to which you can expand a topology to guarantee reasonable effort, consistent performance and maintaining topology typical properties after enlargement.

Connectivity

The connectivity is the minimum number of nodes or links (edge ​​or node connectivity) that need to be cut so that the network is no longer functional as such. It is a measure of the number of independent paths, which may be between two different nodes. This also describes the resilience of the network, ie the higher the connectivity, the fail-safe is the network.

Physical Topologies

Point-to -point topology

An often overlooked because of their simplicity, but the basic topology is point-to -point topology or two-point topology. It comes into being when two nodes are directly connected. All complex topologies based on this simple design principle. Two-point topologies are due to the direct connection to the most powerful constructs and found still partially use, eg in the field of fiber channel networks. A fully permuted point-to -point connection of all nodes in a system is a full intermeshing.

Benefits

• This network provides the highest transmission rate
• Fail-Safe
• Bugproof
• Predictable, usable transmission capacity
• Easily expandable
• Easily understandable
• Easy Troubleshooting
• No routing required

Disadvantages

• Registration of each user on each computer
• No central administration possible
• Shares will not be possible at the user level

Star topology

In networks in a star topology, all other participants are to a central station is connected with a two-point connection (see also Stern graph ). The central subscriber does not necessarily have a special control intelligence. In the transport network that is generally not the case. In computer networks, it may be a specialized facility, such as a hub or switch. Also, a PBX is usually designed as a star network: the switching system is the central node to which the user sets are connected in a star shape. In any case, a central component causes in a network a higher probability of failure for each connection: a failure of the central participant inevitably causes the loss of all connection options at the same time. A common protective measure in star networks is the key component to double (redundancy).

Benefits

• The failure of a terminal device does not have any impact on the rest of the network.
• This network offers high transfer rates when the network node is a switch.
• Easily expandable
• Easily understandable
• Easy Troubleshooting
• Combined telephone / computer network wiring possible
• Very suitable for Multicast-/Broadcastanwendungen
• No routing required

Disadvantages

• Due to failure of the distributor network traffic is impossible
• Low transmission rates with many hosts if a hub is used → dividing the network with a switch is necessary

Examples

• Telephone network
• Fast Ethernet ( physical)
• Token Ring ( physically )

Ring topology

During the crosslinking in a ring topology, two subscribers are connected to each other via two-point connections, so that a closed ring is formed. The information to be transmitted is passed from participant to participant until it reaches its destination. To avoid duplication, this type of networking special addressing methods are needed. Since each participant can act as a repeater at the same time (if no chips are used ), can in this way great distances to be bridged (using fiber-optic cables (FO ) in the kilometer range).

In case of failure of one of the participants the entire network collapses, unless the participants mastered protection switchover. In a ring with protection of the working path is frequently performed in a specific direction of rotation of the ring ( for example, clockwise), the backup path in the other rotational direction ( counterclockwise in this example ). Using this method finds among other fieldbus systems on optical fiber base.

If a multistation access unit ( German: RLV, English: MAU Media Access Unit ) is used, so that the failure of the entire network is prevented in case of failure of a terminal. Each device is connected to only one cable to the RLV. The RLV passes the data here from one port to the next. This one has technically a star, but a logical ring topology. Also the connection of several RLV is possible, with the ring topology is maintained.

A special form of a ring topology, the line topology, in which there is a " ring-opening " that is the first and last machines are not connected to each other. This system is very easy to set up, but also very vulnerable because of the failure of one computer makes all further communications are impossible.

Data of the ( obsolete ) IBM Token-Ring:

• Maximum length 800 m ring
• Computer can be up to 100 m away from the MAU
• Transmission rate of 4, 16 or 100 Mbit / s
• Active topology
• Transport protocol is token passing
• Access is deterministic ( certain access )
• Is addressed only by MAC address

Benefits

• Deterministic computer network communication without packet collisions - predecessor and successor are defined
• All stations operate as an amplifier
• All computers have equal access to opportunities
• Guaranteed bandwidth
• Scales very good degree remains constant during expansion
• Regular topology, therefore, easily programmable

Disadvantages

• Low Bisektionsweite and connectivity, ie the one hand, that the failure of a terminal causes the entire network communication is interrupted (except for protection switching - see: FDDI). That's true for new maps but no longer, since each card has mastered this protection switchover. On the other hand, there are few alternative routes, which in the case of high load conditions can lead to a ring portion bottlenecks
• Expensive components
• May / may not be used for combined Rechnernetz-/Telefonverkabelung
• Data transmissions can be easily intercepted.

Examples

• Token Ring (logically)
• FDDI ( physically )

In theory, one often sees that the physical ring structure follows the logic of order and thus save costs lead lengths, but this is usually done at the expense of flexibility for extensions.

Bus topology

In a bus topology, all devices are directly connected to the same transmission medium, the bus. There are no active components between the devices and the media. The transmission medium is often implemented directly on a circuit board for systems with a smaller physical size, and otherwise as a cable or cable bundle. An example of a network with a bus topology is 10 Mbit / s Ethernet. In the variant Thin Ethernet, there is a single cable, which is divided into segments. The connection between the devices (ie network cards) is done and the segments of the cable over tees. Terminating resistors at the ends of the cable are used to prevent reflections.

In the bus topology has to be ensured that always sends signals to the transmission medium, only one device at the same time. This can be controlled by a central unit, the so-called bus arbiter. Before a device is allowed to send, it must provide a separate line a request to the bus arbiter. And time-sharing method can be used. A central control but is often impractical especially in dynamic networks such as computer networks. Therefore, concurrent accesses the data in these networks detected and resolved the resulting problems. An often used method is for example CSMA / CD.

Benefits

• The failure of a device has the functionality of the network are no consequences
• Very cost-effective, as only small quantities of cables are required
• Simple cabling and network expansion
• There are no active network components needed

Disadvantages

• Data transmissions can be easily intercepted (keyword: sniffer).
• A disturbance of the transmission medium at a single location in the bus ( defective cable ) blocks all together.
• It can at any time always only one station to send data. Meanwhile, all other channels are blocked ( need to transfer data between internal buffer ).
• For buses that allow collisions and putting on a retrospective remedy, the medium can be utilized only to a small extent, because a disproportionate number of collisions occur at higher traffic.

Examples

Tree topology

Tree topologies are characterized in that they (the first or top nodes) have a root, out of which one or more edges (links). This continues to lead to a leaf ( terminal node ) or recursively to roots of other trees (see tree (graph theory ) ). Technically, the tree topology is a network topology in which multiple networks are hierarchical star topology connected. These connections between the distribution ( hub, switch ) via an uplink must be established. Frequently, this topology is used in large buildings.

Benefits

• The failure of a terminal has no consequences
• Structural extensibility
• Long distances achievable ( combination )
• Well suited for searching and sorting algorithms

Disadvantages

• If one distributor ( root ) is the whole of which outgoing (sub ) tree of the distributor no longer available
• Towards the root, it can lead to bottlenecks caused by the defined for trees Bisektionsweite of 1, since a lower tree half must be gone in the other half always on the root to the communication of the
• Trees have with increasing depth ( = number of outgoing links from the root to a leaf ) a very high diameter. This results in connection with the Bisektionsweite poor latency characteristics for classical trees

To counteract these quite serious disadvantages in practice, a plurality of tree variations may be used.

K- tree

The k- tree has come a classic tree from each root go but k edges. Thus one can achieve a shallower depth and therefore lower latency, for example, compared to binary trees. However, a disadvantage is the increased complexity of the root elements ( degree k ).

Ring Extended Tree

A ring- extended tree is a regular binary tree or K, the sheets were, however, coupled to each of the same plane to form a ring ( a so-called horizontal rings). Here you can either leaves all levels coupled into rings, or only the certain (usually deeper ) levels. This leads to a reduction of the root elements upper levels, since nodes in a level now virtually able to communicate locally, without having to a few levels and then go up again down. In practice, typically only a few coupled node of a level (for example the two outermost and the middle ), a so-called interrupted annular. This one has the advantage that it is less expensive than a complete ring, but it still partly above mentioned offers advantages. He is virtually a compromise solution.

Hypertree

The hypertree works on the same principle as the ring-expanded tree, the additional connections but are not limited to the horizontal, but connect nodes of different levels together. However, this requires a relatively complex routing.

Big tree

The fats tree or english fat tree tries to solve the problem of low Bisektionsweite. This is achieved by an increased bandwidth in the direction of the root, for instance by a plurality of parallel links from the root node to the lower levels. This fixes the disadvantage that the root of the tree may become a bottleneck, the high diameter of a tree can be unaffected.

Meshed network

In a mesh network, each terminal is connected to one or more other terminals. If each participant is connected to every other participant, one speaks of a fully meshed network.

In case of failure of a terminal or a pipe it is usually possible, the data to communicate by routing (routing) on.

Benefits

• Secure version of a computer network
• If one terminal data communication is still possible by redirecting (high connectivity)
• Very powerful high Bisektionsweite, base diameter ( at fully meshed networks constant at 1)
• Fully meshed networks require no routing, since there are only direct

Disadvantages

• Lots of cable is necessary; even when not fully meshed computer networks very complicated (usually high level )
• Comparatively complex routing for not fully meshed networks because they are not needed then regularly and not symmetrical, causing many special cases

Cell topology

The cell topology is mainly used in wireless networks to use. A cell is the area around a base station ( for example, wireless access point ) in which a communication between the terminals and the base station is possible.

Benefits

• No cables needed
• No interference due to failure of devices

Disadvantages

• Extremely susceptible to interference and limited range
• Very unsafe because anyone can access it from the outside (encryption required)

Examples

• IEEE 802.11 (Wireless LAN)
• GSM
• Bluetooth

Hybrid topologies

Hybrid topologies use two topologies in a network.

Star - bus

A star bus network occurs when different distributor respectively form the center of a star, this distribution, however, are connected by a bus cable. This variation is often used for buildings with several floors.

Star-star

A star-star network arises when different distributors, each the center of a star form and this distribution in turn are connected via separate cable with a distributor, this variant also likes to come in multi-storey buildings for use.

Logical topology

The logical topology of computer networks can be different from the physical. So Ethernet can be constructed physically as a star or (obsolete) as a bus - are logically necessary to distinguish here at the coupling component used. A hub, there is a logical bus topology, because of the flow of data from a terminal occurs simultaneously with all other devices. However, using a switch, the logical topology is a star, or a point to point connection. However, an exception is broadcast traffic; here also makes the switch logically like a bus because it forwards the data to all connected devices. Token Ring is implemented physically as a star on a multistation access unit ( MSAU ), but is a logical ring topology, since logically the data flow runs from terminal to terminal. ARCNET is built physically as a tree over several active and passive hubs, data flow takes place but also from terminal to terminal and is thus logically a ring topology. The logical topology of a wireless network is the bus topology. ( See also VLAN).

The logical topologies are also so-called overlay networks that have gained particular by the popular peer-to -peer networks ( P2P networks Abbr ) in importance. Overlay networks usually form logical network structures based on minor physical structures. In this case, the topology of the overlay network may be completely different from the topology of the underlying physical network. For example, many structured P2P networks on tree or ring topologies, although the underlying physical structures traditionally follow a star topology.