Network performance

Network performance

Performance is more than speed

Measure	Meaning	Effect
Bandwidth	The theoretical capacity of a link per unit time	Limits possible transfer rate
Throughput	The actual rate achieved	Determines how quickly useful data arrives
Latency	Delay between sending and receiving	Makes interaction feel slow
Jitter	Variation in packet delay	Disrupts real-time audio and video
Packet loss	Packets fail to reach the destination	Causes gaps or retransmission

Match the evidence

Determine the main issue in each case.

1. A 100 Mbps link transfers at 42 Mbps.
2. Video-call packets arrive at irregular intervals.
3. A button response takes 600 ms even though the message is tiny.
4. Some voice packets never arrive.

Check your answer

1. Throughput is below bandwidth.
2. Jitter.
3. Latency.
4. Packet loss.

Calculate transfer time carefully

Use compatible units and distinguish bits from bytes.

transfer time (seconds) = data size (bits) / throughput (bits per second)

A 20 megabyte file contains approximately 160 megabits. At an actual throughput of 40 megabits per second, the idealised transfer time is:

160 Mb / 40 Mb/s = 4 seconds

Real transfers may take longer because of protocol overhead, changing conditions, competing traffic and retransmissions.

Calculate a transfer

A 75 MB video transfers at 30 Mbps. Determine the idealised time.

Check your answer

75 MB × 8 = 600 Mb; 600 / 30 = 20 seconds.

Packet switching

Packet switching divides data into packets that can travel through a shared network. Each packet carries control information used to deliver and reassemble the data.

1. The sender divides and labels the data.
2. Network devices forward packets towards the destination.
3. Packets may take different routes or arrive out of order.
4. The receiver checks and reassembles them.
5. Missing or damaged data may be recovered according to the protocol.

Sharing links improves network use, but variable routes and congestion can change delay and loss.

Routing and forwarding

• Routing determines paths and maintains information about how networks can be reached.
• Forwarding is the local action of moving an arriving packet to the appropriate outgoing interface.

A route may change when a link fails or another path becomes preferable. More hops do not automatically prove poor performance; link quality, distance, load and processing also matter.

Explain the relationship

Explain how routing and forwarding work together.

Check your answer

Routing creates or updates knowledge about reachable paths. A router then uses that knowledge when forwarding each packet to its next hop. Routing informs the decision; forwarding performs it.

Congestion and flow control

Congestion occurs when offered traffic exceeds the capacity of network resources. Queues grow, latency rises and buffers may discard packets.

Congestion control adjusts sending behaviour in response to network conditions. Flow control prevents a sender from overwhelming the receiving endpoint's ability to process data.

Read a performance pattern

Time	Throughput	Latency	Loss
10:00	82 Mbps	24 ms	0%
12:30	18 Mbps	310 ms	6%

Analyse the likely change at 12:30.

View a sample solution

The combination of reduced throughput, sharply increased latency and packet loss is consistent with congestion or another serious link-quality problem at 12:30. More evidence is needed to locate the cause.

Error detection and correction

Transmission systems use error-control processes to detect corrupted or missing data and respond appropriately.

• A checksum or other integrity check can reveal changed data.
• Sequence information can reveal missing or reordered segments.
• Acknowledgments confirm receipt.
• Retransmission replaces missing or damaged data.
• Some systems use forward error correction to recover certain errors without retransmission.

Retransmission supports reliability but consumes capacity and adds delay. Real-time media may prefer a small gap over waiting for a late packet.

TCP, UDP and application needs

Transport	Useful characteristics	Typical priority
TCP	Connection-oriented, ordered reliable byte stream with flow and congestion control	Complete, ordered delivery
UDP	Connectionless datagrams with low transport overhead and no built-in delivery guarantee	Timeliness or application-controlled recovery

Applications may add their own reliability or security. “UDP is always faster†is too broad; suitability depends on implementation and purpose.

Quality of service

Quality of service (QoS) mechanisms classify and manage traffic so important flows receive appropriate treatment during contention.

• Prioritise delay-sensitive voice over a background software download.
• Reserve or shape traffic to protect essential services.
• Manage queues to reduce delay and loss for selected classes.

QoS manages limited capacity; it does not create unlimited bandwidth. Prioritising one class may delay another, so priorities must reflect actual requirements.

Choose a priority

A hospital link carries a live remote consultation and a scheduled archive backup. Recommend a QoS priority and Explain the trade-off.

Check your answer

Prioritise the real-time consultation because latency, jitter and loss directly affect communication needed now. Shape or schedule the backup with lower priority; it may finish later but its complete data can still be transferred reliably.

Evaluate with application criteria

Use the required task to judge measurements.

Application	Especially important
Video call	Low latency, jitter and loss; adequate throughput
Large backup	Sustained throughput and reliable complete delivery
Emergency alert	Low delay, reliable delivery and availability
Interactive game	Low latency and jitter

Independent practice: community event network

A temporary event network supports payment terminals, a live stream and overnight photo uploads. At peak time, measurements show 95% link utilisation, latency rising from 28 ms to 420 ms, 4% packet loss and highly variable video delay. Payments sometimes time out.

1. Analyse the evidence to identify the likely performance condition.
2. Explain how it affects payments and live video differently.
3. Recommend two refinements, including one QoS decision.
4. Evaluate the claim: “The only solution is to increase bandwidth.â€

View a sample solution

1. Very high utilisation with rising latency, loss and variable delay strongly indicates congestion.
2. Payment requests may time out and retry, risking delay or duplicates; live video suffers from jitter, gaps and reduced quality.
3. Prioritise small payment traffic and delay-tolerant control traffic appropriately, while shaping or scheduling photo uploads outside peak time. Additional capacity may also help if monitoring confirms sustained saturation.
4. The claim is too absolute. More bandwidth could reduce saturation, but traffic scheduling and QoS address competing priorities, and client/server timeout or duplicate-handling problems may also need refinement.