Linear topology (daisy-chaining)

A linear topology is exactly what it sounds like – a line! It is the most primitive extension of the point-to-point topology, one of the simplest logical topologies conceptually, and often one of the cheapest in terms of a corresponding physical implementation. In a linear network topology, we extend our point-to-point model in such a way as to only ever have one node connected to at most two other nodes at a given time. The benefit here is obviously in the physical implementation cost (even with high resiliency, this configuration can only ever get so complicated). The drawback, however, should be similarly obvious. Communication from one node to any node other than one of its nearest neighbors will require the intermediary nodes to do some work investigating the target of the inbound request and determine if they are suitable to process the request, and if not, know to pass the request along to the neighbor that didn't originate the request.

Note that specification to not return the request to the neighbor that originated the request in the first place. If the nodes responded by simply submitting the request back out to both of your neighbors blindly, you would end up in an infinite loop of submitting and re-submitting the request between the two nodes. At least one of the nodes in any pair has to be aware enough to not re-submit a request to its originator. This highlights the most important drawback of this topology. Specifically, that it requires a tight coupling of nodes to their conceptual position in the network structure. 

While none of this is particularly complicated, you can already see how the logical organization of your network can impact the design of your networking code. That will only become more apparent as the complexity of the topology increases.