Part 1 of this FAQ started out with conceptually simple, widely used, and indispensable point-to-point and bus topologies often implemented using RS-232, RS-485, and Ethernet standards. These basic topologies are suitable in many situations but are inadequate as the number of nodes increases, or as other performance issues become priorities. Part 2 looks at other network approaches.

Q: What is a ring topology?

In a ring topology, each node is connected only to adjacent nodes on both sides. Messages are passed from originating node to target node by going through each intermediate node along the way, Figure 1. The network is managed either by use of precise time slots (time division multiplexing) or by passing a “token” from node to node: only the node with the token can transmit, thus avoiding collisions. Also, the maximum time for a message to reach the target node is deterministic, based on the maximum possible number of intervening nodes between originating and target nodes.

The ring does have weaknesses: each node, and each link between nodes, is a potential failure point for the entire system, and any break could compromise the ring completely or in part, depending on where it occurs. To overcome this, many rings allow passing of tokens in both directions (counter-rotating tokens) which can also speed up performance under non-failure conditions. Also, each node of the ring is busy passing tokens and so has reduced resources remaining for other functions.

Q: What about the star configuration?

A: In the star approach, there is a central node called a hub, and then a separate point-to-point connection from the central hub to each of the other nodes, Figure 2. All data traffic from one node to another goes through the hub. This can be efficient in terms of use of resources and bandwidth compared to the ring. However, it costs a great deal in cabling (if wired) and RF channels (if wireless) while the central hub is a single point of failure which would bring down the entire network if it fails or is compromised. In some applications, though, the star with its central hub is a good fit in terms of cost, performance, and reliability.

Q: Are there other network topologies?

A: Of course! The mesh network is a general term covering a variety of interconnection approaches. In a basic mesh network, each node is connected to other nodes using point-to-point links, Figure 3. In a fully connected mesh, each node connects to all other nodes in the system. This is efficient for throughput and efficiency but obviously become unwieldy as the number of nodes grows and the distances increase. In a partially connected mesh, each node is connected to a few nearby nodes, Figure 4, which reduces interconnect complexity significantly as the number of nodes increases.

The mesh as two advantages for a network with many nodes. It provides some amount of network resiliency, since there are multiple paths between any one node and another one located further away. Also, it is attractive because each node only needs enough power or circuitry to reach its nearby connected nodes, which is a virtue wireless or battery-powered systems.

However, the mesh approach means that intermediate nodes may be using their bandwidth and capacity to pass-through messages between other nodes, so the overall power requirements, performance, and throughput are harder to assess in advance.

Q: Are we done?

A: No. There is also a tree topology, which uses a combination of star networks, connected to each other via a bus, Figure 5. Each node in this hybrid approach is called a leaf, and if a leaf hub faults, only its members are lost as long as the failure does not short the entire bus, or go into non-stop “chatter” mode (there are ways of preventing either of those fault consequences, through isolation and watch-dog timers.

Q: What’s the difference between physical versus logical network topology?

A: The physical topology is what a wiring or configuration schematic shows as the interconnect arrangement. In contrast, the logical topology relates to how the nodes are addressed and managed, how they interact with each other, the allowed message paths, and other “soft” issues. It is entirely possible or example, to have a bus network which is managed like a star, confusing as that may seem. Network management is a large and separate topic.

Q: So, what’s the best or “right” network topology?

A: As in almost all engineering decisions, there is no single “best,” as the choice depends on the specifics of the application and situation, weighing factors such as cost, distance, ruggedness, reliability, latency, deterministic or random performance, resilience, throughput, management complexity, power requirements, and many other factors. The designer must prioritize the project objectives as the basis for the analysis. Keep in mind that many “best fit” networks are actually a hybrid blend of one or more of the basic topologies. Often, this provides the right balance among the many conflicting requirements, even at the cost of more complicated network management issues and protocols.

References

Lucid Software, Inc., “What is a Network Diagram”

Computer Hope, “Computer terms, dictionary, and glossary”