中文（繁體）
中文（简体）
English
DSE ICT | Cloud Computing and IoT Lesson-Prep Guide (Tutor Edition) — IaaS, PaaS and SaaS, Data Volume Arithmetic and Edge Trade-offs
Students say a system is in the cloud but cannot say which layer it sells, nor how much data a thousand sensors make. This lesson turns the vocabulary into three calculations.
Objectives — Students can: (1) separate IaaS, PaaS and SaaS by asking which layer the user still manages, and pick a model for a given scenario with a reason; (2) compute daily and monthly data volume from sensor count, sampling rate and record size, then convert it to bandwidth and cost; (3) compare edge and cloud latency, decide which applications must sit at the edge, and state the cost of that choice.
01 Prerequisites (3-minute check)
If any of these fails, fix it before the lesson starts.
| Skill | Quick check | Pass mark |
|---|---|---|
| Unit conversion | How many megabytes in a gigabyte | 1,024 MB, or 1,000 MB when stated |
| Bits and bytes | How many bits in a byte | 8 bits |
| Ratios | How many times larger is 120 ms than 15 ms | 8 times |
02 Lesson flow (60 minutes)
| Time | Stage | Tutor does | Student does |
|---|---|---|---|
| 0-10 min | Three service models | Draw one line for who manages what | Say what the user still runs |
| 10-22 min | Four IoT layers | Sensing, network, platform, application | Place each device |
| 22-36 min | Data volume | Count x rate x size x days | Reach 53.1 GB |
| 36-48 min | Latency | Edge 15 ms against cloud 120 ms | A cut of 87.5% |
| 48-56 min | Availability and cost | 99.9% against 99.99% downtime | 8.76 hours a year |
| 56-60 min | Exit check | Three questions marked on the spot | All three right |
03 Teaching script (read it out)
Scene 1: Cloud sells the layer you no longer manage
The three models differ on one question: how far up the stack the user is responsible. IaaS gives virtual machines and storage, so everything from the operating system upward is yours; PaaS gives a runtime and database, so you only ship code; SaaS gives finished software, so you manage accounts and permissions.
Choosing is a trade between control and staffing: custom kernel settings mean IaaS, code-only deployment means PaaS, and email or forms mean SaaS. A school on Google Workspace is buying SaaS, a company running its own virtual server is buying IaaS, and a team shipping an API to a managed platform is buying PaaS.
The common lost mark is treating cloud as one product rather than a stack.
Choosing is a trade between control and staffing: custom kernel settings mean IaaS, code-only deployment means PaaS, and email or forms mean SaaS. A school on Google Workspace is buying SaaS, a company running its own virtual server is buying IaaS, and a team shipping an API to a managed platform is buying PaaS.
The common lost mark is treating cloud as one product rather than a stack.
Ask students: in SaaS, who patches the operating system (the provider, not the user).
Scene 2: Every IoT question starts with a chain of multiplication
A smart campus has 1,200 sensors, each reading every 15 seconds, with 256 bytes per record.
Readings per sensor per day: 86,400 / 15 = 5,760, so 1,200 of them give 6,912,000 records; at 256 bytes that is 1,769,472,000 bytes, about 1.769 GB a day, and 53.1 GB a month.
As bandwidth that is only 1.769 GB over 86,400 seconds, about 20.5 kB/s, which is tiny; storage and queries are what accumulate, 645.9 GB a year. At 0.023 US dollars a gigabyte the monthly bill is 1.22 dollars, so the real cost driver is usually queries and requests, not capacity.
Compressing 4:1 first cuts the month to 13.3 GB, the standard move of reducing before storing.
Readings per sensor per day: 86,400 / 15 = 5,760, so 1,200 of them give 6,912,000 records; at 256 bytes that is 1,769,472,000 bytes, about 1.769 GB a day, and 53.1 GB a month.
As bandwidth that is only 1.769 GB over 86,400 seconds, about 20.5 kB/s, which is tiny; storage and queries are what accumulate, 645.9 GB a year. At 0.023 US dollars a gigabyte the monthly bill is 1.22 dollars, so the real cost driver is usually queries and requests, not capacity.
Compressing 4:1 first cuts the month to 13.3 GB, the standard move of reducing before storing.
Ask students: why 1,200 sensors still average only 20 kB/s (the sampling is sparse).
Scene 3: Latency decides where the computing sits
The same decision takes 15 ms at the edge but 120 ms going to the cloud and back, 8 times longer, a 87.5% cut in delay.
Automatic braking, robot arms and heart monitoring need an answer inside 50 ms and must run at the edge; monthly reports, model training and long-term trends belong in the cloud.
State the price: edge devices need computing power and electricity, firmware updates on every unit, and more security boundaries; the cloud is centrally managed and elastic but depends on the network.
Availability also needs arithmetic: 99.9% is 8.76 hours of downtime a year, 99.99% is 52.6 minutes; homework systems can live with 99.9%, payments and medical systems need 99.99% plus failover. IPv6 offers about 3.4 x 10 to the 38 addresses, which is what makes addressing every sensor possible.
Automatic braking, robot arms and heart monitoring need an answer inside 50 ms and must run at the edge; monthly reports, model training and long-term trends belong in the cloud.
State the price: edge devices need computing power and electricity, firmware updates on every unit, and more security boundaries; the cloud is centrally managed and elastic but depends on the network.
Availability also needs arithmetic: 99.9% is 8.76 hours of downtime a year, 99.99% is 52.6 minutes; homework systems can live with 99.9%, payments and medical systems need 99.99% plus failover. IPv6 offers about 3.4 x 10 to the 38 addresses, which is what makes addressing every sensor possible.
Ask students: why automatic braking cannot rely on the cloud alone (network delay and disconnection).
04 Question types and worked steps
| Type | Example | Steps | Answer |
|---|---|---|---|
| Classification | IaaS, PaaS, SaaS | Ask which layer the user manages | SaaS leaves accounts only |
| Calculation | Monthly data volume | Count x rate x record size x days | 53.1 GB |
| Calculation | Latency reduction | (old – new) / old x 100% | 87.5% |
| Essay | Edge or cloud | Latency need, bandwidth, power, management | Split by application |
Sanity check (teach this): Convert the sampling rate to a daily count first: once every 15 seconds is 86,400 / 15 = 5,760, not 15.
Keep bits and bytes apart: 256 bytes is 2,048 bits.
A gigabyte here is a thousand million bytes, and the 1,024 convention is used only when the question says so; never mix them.
A latency cut is relative to the old value, so the new value is not 87.5% of the old.
Keep bits and bytes apart: 256 bytes is 2,048 bits.
A gigabyte here is a thousand million bytes, and the 1,024 convention is used only when the question says so; never mix them.
A latency cut is relative to the old value, so the new value is not 87.5% of the old.
05 Independent practice (3 questions)
| Question | Answer | What to watch |
|---|---|---|
| Daily records for 1,200 sensors every 15 seconds | 6,912,000 records | 86,400 / 15, then x 1,200 |
| 30-day total at 256 bytes each | 53.1 GB | Multiply, then divide by a thousand million |
| Latency cut from 120 ms to 15 ms | 87.5% | 105 divided by 120 |
06 Common mistakes and what to say
| Mistake | Student writes | Say this |
|---|---|---|
| Taking the sampling interval as the daily count | Answering 15 records a day | Once every 15 seconds means dividing 86,400 by 15. |
| Mixing bits and bytes | Reporting eight times the volume | One byte is eight bits. |
| Treating cloud as a single product | Answering that cloud just means online | Name the layer: IaaS, PaaS or SaaS. |
| Reporting the new value as the cut | Answering that latency is 87.5% | The cut is relative; the new value is 15 ms. |
07 Homework (paper, section and question numbers)
- Secondary ICT chapter index — All chapters and practice
- All lesson-prep guides — Compare other subjects
08 Exit check (last 5 minutes)
| Question | Answer | Decision |
|---|---|---|
| Daily records for 1,200 sensors | 6,912,000 | All three right: move to cloud security |
| Why 53.1 GB a month | 1.769 GB a day times 30 | One wrong: reteach the chain |
| Latency cut from 120 ms to 15 ms | 87.5% | Two wrong: return to act three |
09 Tutor FAQ
Must students memorise provider names?
No. The exam tests layers and trade-offs, so school or home examples are enough.
Does IoT always need the cloud?
No. A small system can use a local server; the cloud buys elasticity and off-site backup.
Will edge computing replace the cloud?
No. They divide the work: the edge reacts instantly, the cloud aggregates, trains and stores.
Find a tutor — freeTutor Registration →
A TutorZone lesson-prep guide. Written by the TutorZone team; questions and lesson flow are original and are not taken from any school or publisher material.
