By Jeff Kabachinski, MS-T, BS-ETE, MCNE
The final installment of our packet review series involves layer 4 of the Open Systems Interconnection (OSI) model. September’s Networking column delved into the Transmission Control Protocol’s (TCP’s) function—also a layer 4 protocol—but wait: There’s another method to transfer network data.
Recall that TCP is called “connection-oriented,” due to the handshaking process that takes place. It’s also referred to as “reliable data-passing,” as individual packets are tagged and counted. User Datagram Protocol, aka: UDP, is called connectionless, however, since there is no handshaking; this, then, also makes it known as “unreliable data passing.”
UDP is one of the protocols within the TCP/Internet Protocol (IP) suite of 200-plus applications or programs, which also include Hypertext Transfer Protocol,(HTTP) and File Transfer Protocol (FTP).
UDP at the transport layer uses the same address allocations as TCP port numbers. Following the port addresses, the data length field follows in the packet structure to indicate the amount of data in the payload (Figure 1), and the packet also includes a header checksum. You’ll notice that there are no handshaking, sequencing, or acknowledgement fields like with TCP—which is why UDP is called “connectionless” or “unreliable.”
The Connectionless Routine
The connectionless routine tries to initiate communication via UDP packets—UDP doesn’t want to take time to see if everything’s been received by counting the number packets sent. It could be due to “real time” information, such as a telephone call.
Voice packets are continually sent—and if one is dropped or missed one once in a while, it only briefly breaks the call. Or, in a clinical setting UDP could be to transmit live electrocardiography (ECG) information to a central monitoring station from a patient monitor. After all, the bedside monitor routinely sends UDP packets of real-time ECG waveform sections to the nurses’ station.
Again, due to using UDP without counting packets, a packet may get dropped occasionally. And dropped packets will show up as a short gap in the patient’s ECG waveform on the nurse’s screen. Note: There’s no need to add to network traffic by counting packets because, by that time, the waveform is already on the way to scroll off the nurse’s screen. Plus, typical central station monitors only show a few seconds of a patient’s ECG waveform. At the bedside monitor, it will be stored to get the entire ECG data, if needed.
But for a so-called “real-time” data network, there’s a need to be cautious about the amount of traffic consuming bandwidth. Using Ethernet to merge into high-network traffic becomes more complicated with high traffic and may delay overall network communication.
On the other hand, when printing out a bedside monitor’s full disclosure of 24 hours of ECG waveforms, you must make sure that every packet is received and sequenced correctly so that the report is complete—lacking any waveform gaps. This also avoids having to print half pages or dotted line waveforms due to lost packets. Therefore, in this case, the network will use TCP—the connection-oriented or “reliable data-transfer protocol.”
Summing It Up
This wraps up our series on packet architecture. The discussion started with an overview of the OSI model, showing how packets are built and disassembled. Ethernet addressing and function were then explored, followed by Internet Protocol version 4 (IPv4) addressing and function. Finally, we delved into the transport layer protocols in terms of TCP (a connection-oriented routine) and UDP (a connectionless routine).
A few notes from the packet series:
- Use the OSI model as a protocol viewer. If you come across a protocol you haven’t seen before and can determine the OSI layer that it resides on, knowing what happens at that layer will clue you into what sort of process the protocol provides.
- Use the mnemonic “Please Do Not Throw Sausage Pizza Away”—PDNTSPA—to remember the OSI layers from the bottom up: the physical, data link, network, transport, presentation, session, and application layers.
- An Ethernet local area network address (six bytes of hexadecimal information where a colon delimits each byte) is also known as the hardware address and the media access control (MAC) address. Ethernet and the MAC sub-layer reside at layer 2 of the OSI model: the data link layer.
- IPv4 is a layer 3 wide area network-addressing scheme. The 4-byte decimal address is four bytes in length—with each byte delimited by a period, or “dot”. It is divided into the network identification portion and the node identification portion as defined by the value of the first byte.
Finally, check out Newton’s Telecom Dictionary: Telecommunications, Networking, Information Technologies, The Internet, Wired, Wireless, Satellites and Fiber, which is a great resource if you need to bone up on IT-related acronyms or networking topics in a hurry. (After all, sometimes all you need to know is what the acronym stands for!.
Jeff Kabachinski, MS-T, BS-ETE, MCNE, has more than 20 years of experience as an organizational development and training professional.