| 3.1 | Network Services | ||
| 3.1.3 | Connectionless services |
|
With connectionless services, there is
no connection set up between the two communicating protocol
entities. This is illustrated in the figure to the left. Each data
unit is transmitted independently of previous and subsequent data
units. Examples of connectionless services include IP at Layer 3 of
the Open System Interconnection (OSI) model and User Datagram
Protocol (UDP) at Layer 4 of the OSI model.
A connectionless service can be compared to mailing a bunch of holiday postcards. Each postcard is handled separately. In other words, each packet (postcard in this example) routes independently. In the data world, this means that a host can transmit data to another host in an unplanned fashion and without prior coordination. You could also compare a connectionless stream to the cascading balls in a Pachinko game (for example, sabgames.com/arcade/pachinko.html), where all the balls end up in the same place but the paths vary widely. When using connectionless protocols, the receiving host that is part of the OSI model protocol stack sees that the packets received are complete, but a higher-level application is responsible for putting them in the proper order and requesting retransmission of any missing packets. As you can probably tell, connectionless data transfer is quite efficient and network traffic demand is entirely a function of how much data is being sent (no overhead). Connectionless service offers two important advantages over connection-oriented service: dynamic path selection and dynamic bandwidth allocation. Dynamic path selection enables traffic to be routed around network failures because paths are selected on a per-packet basis. With dynamic bandwidth allocation, bandwidth is used more efficiently because network resources are not allocated a bandwidth that they will not use. When troubleshooting connectionless data transfers, look for problems where data is not acknowledged by an upper layer protocol or application, where errors in the data are not reported to the sender, where data may arrive out of order, and where there is no flow control. When connectionless protocols are implemented at the lower layers of the OSI model, the upper layers are usually connection oriented to ensure the reliable transfer of data. A common example is the use of IP at Layer 3 and TCP at Layer 4. Connectionless services are useful for transmitting data from applications that can tolerate delay and resequencing. Database applications typically rely on connectionless service, as do DNS, TFTP, NFS, SNMP, and NTP.
|