Services

Computer Network & Network Security System — Data Link Layer, NEC licence examination syllabus (Nepal Engineering Council).

Data Link Layer Services: the job description

Before diving into CRC, Hamming code, and sliding windows, get clear on WHY this layer exists at all.

Imagine two neighbours shouting messages across a noisy street. Some words get lost, some arrive garbled, and sometimes one person talks so fast the other can't keep up. The data link layer's entire job is to fix exactly these problems for two directly-connected devices.

🧰 The Four Core Services

FramingGroup raw bits into manageable, addressed units called frames.

Flow controlMake sure a fast sender doesn't overwhelm a slow receiver.

Error controlDetect (and sometimes fix) frames damaged in transit.

AddressingIdentify which device on the local network a frame is meant for (via MAC address).

LLC and MAC sublayers

The data link layer itself is often split into two sub-layers: LLC (Logical Link Control — handles flow/error control, talks upward to the network layer) and MAC (Media Access Control — handles addressing and who gets to use the shared medium, talks downward to the physical layer).

💡 One-liner: "Why split data link layer into LLC and MAC?" → LLC stays the same regardless of the physical medium; MAC changes depending on whether it's Ethernet, Wi-Fi, etc. — separating them lets one LLC design work across many MAC types.

Each service answers one failure of a raw link

The four services are easier to hold if you ask what goes wrong without each one.

🔧 What a bare physical link cannot do

Without framingThe receiver gets an endless bit stream with no idea where one message stops and the next begins.

Without error controlNoise flips bits and the receiver accepts corrupted data as genuine, never knowing.

Without flow controlA fast sender overruns a slow receiver's buffer and data is silently dropped.

Without access controlOn a shared medium, two stations transmit simultaneously and both signals are destroyed.

The physical layer moves bits and promises nothing about them. Every data link service exists to convert that raw, unreliable bit-pipe into something the network layer can use — a channel that delivers identifiable, checked frames at a rate the receiver can absorb.

Why the same services reappear at the transport layer

Flow control and error control appear in both layers, and questions ask why the duplication is not waste.

DATA LINK between two DIRECTLY connected devices one hop, one wire TRANSPORT between two PROCESSES, end to end across many hops and routers

A perfect data link on every hop still does not give end-to-end reliability. A router in the middle can drop a packet when its buffers overflow, and no link-layer check anywhere sees that — the frame arrived correctly on every wire it travelled. Only the endpoints can tell whether the data actually arrived, which is the end-to-end argument from the transport chapter.

💡 So they are not duplicates. Link-layer error control catches noise on one wire immediately, where retransmission is cheap and local; transport-layer error control catches everything else, including loss inside routers that no link ever sees. Fixing errors early is an optimisation; fixing them end to end is the correctness guarantee.

Connectionless and connection-oriented at this layer

📮 Three service types

Unacknowledged connectionlessSend and hope. No connection, no acknowledgement, no retransmission — right for links with very low error rates or for real-time traffic. This is what Ethernet does.

Acknowledged connectionlessEach frame is acknowledged individually, but there is still no connection setup. Useful on unreliable links such as wireless.

Acknowledged connection-orientedA connection is established, frames are numbered and delivered exactly once and in order — the strongest guarantee and the most overhead.

💡 Ethernet choosing the weakest option is deliberate, not lazy. Wired links are reliable enough that per-frame acknowledgement would cost more than the errors it prevented — so the recovery is left to TCP, which has to be there anyway. Wireless makes the opposite choice, because its error rate is high enough that local retransmission is far cheaper than an end-to-end one.

Syllabus points

Framing, flow control, error control, addressing
LLC and MAC sublayers

Create a free account to tick topics off, take notes as you read, watch the video lessons and get a day-by-day study plan built around your exam date.