S 5.2 Selection of an appropriate network topography

Initiation responsibility: Head of IT Section

Implementation responsibility: Network planner; Head of Site/Bldg Technical Service

The topography of a network is the purely physical structure of the network as it is visible with cables. In contrast to this, the topology of a network is the logical structure as it appears to network components. The topography and topology of a network are therefore not necessarily identical. By nature, topography mostly relates to the spatial environment of the building. These are amongst others:

• locations of the network subscribers

• available space for cable routes and cables ( S 1.21 Sufficient dimensioning of lines)

• required cable types ( S 1.20 Selection of cable types suited in terms of their physical/mechanical properties)

• specifications regarding cable protection ( S 1.22 Physical protection of lines and distributors)

The advantages and disadvantages of various possible topographies are discussed in the following. Other conceivable topographies which are not mentioned in this chapter can be considered as special cases of the structures described here.

In general, a distinction can be made between two basic types of configuration, star and bus, which can also be extended to form the tree and ring configurations respectively. These four types are described briefly in the following:

Star

All subscribers in a star network are linked with a central node via a dedicated line. The cabling of the frequently-used Token-Ring architecture topographically results in a star configuration, but functions topologically as a ring.

Advantages:

• Impairment of a line will only affect the operations of the system connected to it.

• Changes in the allocation of network subscribers to connection points at the central node and separation of individual subscribers can be performed centrally.

• A topographical star configuration can serve as a basis for forming any conceivable topology.

Disadvantages:

• Failure of the central node will result in the failure of all connected IT systems.

• Extensive cabling is required due to the separate linkage of each subscriber to the central node.

• With an increasing number of individual lines, the risk of cross-talk will increase.

• Cabling in star configuration might restrict the communications range, depending on the cable type and communications protocol in use (refer to S. 5.3 Selection of cable types suited in terms of communications technology). Repeaters can be used to solve this problem, although they prove very expensive if a large number of lines is involved. Furthermore, it is not possible to insert any required number of repeaters into a line. The maximum number here also depends on the protocol in use. Another alternative is to convert the network to a tree configuration.

Tree

A tree structure is formed by linking together several star networks. In this case, the network subscribers are assigned to groups which are connected in star configuration to decentral network nodes. These decentral network nodes are linked mutually via one line or several dedicated lines. In certain cases, all the decentral nodes are also routed to one central network node.

Advantages:

• As concerns a linkage of the systems to the decentral network nodes, the same advantages apply here as in the case of the star network.

• For new subscribers, new cabling is required only in the area of the corresponding decentral network node.

• Given an appropriate configuration of the decentral network nodes, an exchange of data between the subscribers of any particular node is possible even if the other nodes fail.

• Connecting the decentral nodes to each other via a single line reduces the cabling requirements.

• Amplification on a single line is sufficient for bridging large distances between the nodes (cost saving).

• It is advisable to link the nodes by means of (usually more expensive) high-quality cables, which can also bridge large distances without the need for additional amplification. As compared with the repeater option which would otherwise be required, this offers advantages in terms of reduced costs and increased reliability.

• A tree structure allows the establishment of redundant links through the meshing of the individual nodes.

Disadvantages:

• Failure of a transition from one decentral network to another will disrupt the operations of all the connected subscribers.

Bus

In the case of a bus, all network subscribers are connected to a common line. This line usually consists of a central cable, to which the individual subscribers are connected via breakout cabling.

Advantages:

• Cabling is reduced to just one cable and any breakout cabling which might be required.

• A subsequent installation of new subscribers generally requires only minimal cabling: the subscribers are simply connected to the existing bus cable.

• A bus can be easily extended through the use of repeaters. However, it must be noted that such extensions are restricted in length in accordance with the type of cable and protocol used (refer to S 5.3 Selection of cable types suited in terms of communications technology).

• Resources can be connected to almost any point on the bus.

• Due to the central cable, the bus cabling takes up much less space than a comparable star configuration with a TP cable.

Disadvantages:

• Interferences in the cable line affect the entire bus.

• An interruption in the bus cable completely paralyses data communications.

• From a certain bus length and number of subscribers onwards, the bus can no longer be extended easily.

• Depending on the type of cable in use, restrictions need to be observed when connecting new subscribers (e.g. the minimum distance between two subscribers).

Ring

Topographically, a ring is a bus whose two ends are connected together. One special type of ring consists of a double-ring like that used with FDDI, for example.

Advantages:

• If a line is interrupted, a ring can continue operation to a limited extent. The type of limitation depends on the network access protocol used for the ring and can involve, for example, losses in bandwidth..

• The optional, double-ring design provides additional redundancy and failure tolerance.

Disadvantages:

• A restricted number of protocols is available for the ring and double-ring systems, i.e. not all protocols can be used on these systems. This can prove disadvantageous for the future development of a network.

Collapsed and distributed backbone

A collapsed backbone is a special variant of a network node whose backplane (a local, high-speed link within a device) incorporates one of the above-mentioned structures or a combination thereof. In the case of a collapsed backbone, all cables are routed centrally to one network node, thus comprising a star configuration in principle. A large variety of structures can be supported within the network node. In the case of a tree structure, for example, very short connection lines within the network node are used to establish the required links between the decentral stars.

Advantages:

• All cable connections can be controlled and managed centrally.

• High transmission rates are generally achieved in the backplane. Depending on the product in use, this results in the availability of the full network bandwidth between the segments. Disadvantages:

• If the collapsed backbone fails, so do all network access ports.

In the case of a distributed backbone, the individual network components belonging to the backbone are spatially distributed and coupled via the standard network infrastructure. Topographical trees, for example, are generally realised by means of a distributed backbone.

As mentioned above, no general recommendations can be made as regards the selection of a suitable network topography. One of the factors which decisively influence any such selection is the structural environment. In general, newly installed networks today are wired in star or tree configuration. Here, it is advisable to use optical fibre cables in the backbone areas (primary and secondary areas) and twisted-pair cables from category 5 or higher for the storey cabling (tertiary area). The primary area contains the cabling which links buildings together, whilst the secondary area contains the cabling which links together the active network components of individual sections within a building (e.g. linkage of individual storeys).

From today's perspective, the selection of these media for the individual areas ensures cabling which has a guaranteed future and which also fulfills high requirements concerning the bandwidth, particularly in the backbone area. In each case however, a check must also be made as to whether a hybrid installation consisting of star and ring configurations is practical or necessary. Here, it is often useful to install primary cabling between buildings as a FDDI double ring, and secondary as well as tertiary cabling in star or tree configuration, as mentioned above.