[Contents] [Back] [<< Prev] [Next >>]

Connection-oriented and Connectionless Modes
The OSI Network layer determines the route a packet will take as it passes through a series of routers from the source PC to the destination PC. The complexity and versatility of Network layer addressing gives rise to two different communication modes for passing messages across the network, both of which are recognized under OSI:
To understand the distinction between connection-oriented and connectionless communications, you must consider an important distinction between the OSI model’s Data Link and Network layers. In theory, the Data Link layer facilitates the transmission of data across a single link between two nodes. The Network layer describes the process of routing a packet through a series of nodes to a destination elsewhere on the network. An example of this latter scenario is a message passing from a PC on one LAN segment through a series of routers to a PC on a distant part of the network. The internal nodes forwarding the packet also forward other packets between other end nodes.

In connection-oriented mode, the chain of links between the source and destination nodes forms a kind of logical pathway— a connection. The nodes forwarding the data packet can track which packet is part of which connection. This enables the internal nodes to provide flow control as the data moves along the path. For example, if an internal node determines that a link is malfunctioning, the node can send a notification message backwards through the path to the source computer. Furthermore, because the internal node distinguishes among individual, concurrent connections in which it participates, this node can transmit (or forward) a “stop sending” message for one of its connections without stopping all communications through the node. Another feature of connection-oriented communication is that internal nodes provide error correction at each link in the chain. Therefore, if a node detects an error, it asks the preceding node to retransmit.

Connectionless mode does not provide these elaborate internal control mechanisms; instead, connectionless mode relegates all error-correcting and retransmitting processes to the source and destination nodes. The end nodes acknowledge the receipt of packets and retransmit if necessary, but internal nodes do not participate in flow control and error correction (other than simply forwarding messages between the end nodes).

The advantage of connectionless mode is that connectionless communications can be processed more quickly and more simply because the internal nodes only forward data and thus don’t have to track connections or provide retransmission or flow control.

Connectionless mode does have its share of disadvantages, however, including the following:
It is important to remember that OSI is not a protocol suite itself—OSI is a standard for designing protocol suites. As such, individual implementations of connectionless protocols can attenuate some of the preceding disadvantages. It is also important to remember that connection-oriented mode, although it places much more emphasis on monitoring errors and controlling traffic, doesn’t always work either. Ultimately, the choice of connection-oriented or connectionless communications mode depends on interoperability with other systems, the premium for speed, and the cost of components.