ELEC3506

Topics

Transport LayerLecture 7 PDF10 min

UDP

What UDP's best-effort service actually gives up, the three advantages that trade-off buys, and the kinds of application that choose it deliberately.

By the end of this page you should be able to

  • List the three characteristics of UDP's best-effort service
  • Give the three advantages UDP has over a connection-oriented protocol
  • Name at least three classes of application that use UDP and say why each one chooses it
  • Explain what the lecture means by UDP adding no value to IP beyond process-to-process communication

The idea

Two protocols answer the connection and reliability questions from the previous topic in opposite ways. UDP takes the connectionless, unreliable corner deliberately — it is built for applications where speed and simplicity matter more than a delivery guarantee.

How it works

Two popular protocols

The lecture names two transport-layer protocols. UDP is connectionless and unreliable, and adds no value to the services of IP except process-to-process communication. TCP is connection-oriented, reliable, delivers in order, and adds congestion control on top of flow control and error control. TCP is the subject of the rest of this module; this topic covers UDP on its own.

How it works

UDP characteristics

  • Connectionless — no handshaking takes place between UDP peers.
  • Each UDP segment is handled independently.
  • Best-effort service — a UDP segment may be lost, or delivered out of order.

How it works

Advantages of UDP

  • Small delay — there is no connection to establish first.
  • Simple — no connection state is kept at the sender or the receiver.
  • Low overhead — the segment header is small.

Aside

The lecture does not give UDP’s header field layout or bit widths anywhere in this material — unlike TCP’s segment format, covered in detail in the next topic. This page does not state UDP header fields for that reason; it was not something the source gave a number for, and nothing here was verified independently to fill the gap.

How it works

Typical applications

UDP suits:

  • Processes needing simple request-response communication, with little concern for flow and error control.
  • Processes that already have their own internal flow-control and error-control mechanisms, so TCP’s version would be redundant.
  • Management processes, such as SNMP (Simple Network Management Protocol).
  • Interactive real-time applications that cannot tolerate uneven delay between parts of a received message.
  • Multicasting.
UDPTCP
ConnectionConnectionlessConnection-oriented
ReliabilityUnreliableReliable
OrderingNot guaranteedIn-order delivery
Extra controlNone beyond process-to-process addressingFlow control, error control, congestion control
The headline contrast. TCP's mechanisms behind each of these get their own topics from here on.

In the exam

  • State all three UDP characteristics as a set: connectionless, independent segment handling, best-effort (lost or out-of-order possible). A list question testing this wants all three.
  • State all three advantages together: small delay, simplicity, low overhead — each tied to a specific cause (no setup, no state, small header).
  • Give the application list by category, not just “streaming video” — simple request-response, self-managed flow/error control, management (SNMP), real-time-intolerant-of-delay, and multicasting.
  • “UDP adds no value to IP except process-to-process communication” is close to an exact quote from the lecture and is worth stating precisely if asked what UDP contributes.

Check yourself

  • UDP is connectionless, handles each segment independently, and offers only best-effort delivery — segments may be lost or arrive out of order.
  • Its three advantages are small delay, simplicity, and low overhead, each a direct consequence of skipping connection setup and state.
  • UDP suits simple request-response traffic, self-managed applications, management protocols like SNMP, delay-intolerant real-time traffic, and multicasting.
  • The lecture does not cover UDP’s header fields — those are not stated here.

Check yourself

  1. A UDP segment is lost in transit. What does UDP itself do about it?
  2. Which of these is one of UDP's three stated advantages?
  3. Why would an application with its own internal flow and error control still choose UDP over TCP?
  4. What does the lecture mean by saying UDP "adds no value to services of IP" beyond process-to-process communication?