106  Cloud Computing for IoT

In 60 Seconds

Cloud computing is renting powerful remote computers on demand, defined by NIST’s five essential characteristics: on-demand self-service, broad network access, resource pooling, rapid elasticity, and measured service. IoT systems use cloud for scalable storage and analytics of massive sensor volumes (millions of readings/day), but edge computing remains essential for real-time control (<100ms latency) and offline operation.

Minimum Viable Understanding

• Cloud computing is renting powerful remote computers over the internet instead of owning your own, defined by NIST through five essential characteristics (on-demand self-service, broad network access, resource pooling, rapid elasticity, measured service).
• IoT systems use cloud for scalable storage, processing, and analytics of massive sensor data volumes, but edge computing is still needed for real-time control and offline operation.
• Cloud evolved from grid computing (scientific HPC) through utility computing (metered resources) to today’s on-demand, elastic, pay-per-use services.

106.1 Learning Objectives

By the end of this chapter, you will be able to:

• Explain Cloud Computing: Define cloud computing using the NIST model and its five essential characteristics
• Trace Cloud Evolution: Explain the progression from grid computing to utility computing to modern cloud
• Articulate IoT-Cloud Benefits: Justify why IoT systems benefit from cloud infrastructure for specific use cases
• Apply NIST Characteristics: Map each NIST characteristic to specific IoT use cases

106.2 Prerequisites

Before diving into this chapter, you should be familiar with:

• Networking Basics: Knowledge of TCP/IP, HTTP, and network protocols is essential for understanding cloud connectivity
• IoT Reference Models: Familiarity with IoT architectural frameworks provides context for how cloud services integrate with device layers

106.3 For Kids: The Giant Computer in the Sky!

Explain Like I’m 5: What is the Cloud?

Have you heard grown-ups talk about “the cloud”? It’s not actually up in the sky with rain clouds!

106.3.1 What IS the Cloud?

The cloud is just really big, powerful computers that live in special buildings (called data centers) far away. When you use “the cloud,” you’re borrowing these super computers through the internet!

106.3.2 A Cloud Story

Imagine you have a tiny toy box at home (that’s your small device). But your toys are getting too many to fit!

Your friend has a HUGE warehouse with endless shelves. They say “You can keep your extra toys at my warehouse! Just tell me when you want to play with them, and I’ll send them to you!”

That’s the cloud! A giant warehouse for your data and computer work.

106.3.3 Why Use Someone Else’s Computer?

At Home	In the Cloud
Your computer might be slow	Super fast computers!
You run out of space	Almost unlimited space!
Costs a lot to buy big computers	Pay only for what you use
If it breaks, you’re stuck	They have backups everywhere!

106.3.4 The Cloud and Your Smart Home

When Temperature Terry reads “75F,” where does that information go?

1. First: Terry sends it through your Wi-Fi
2. Then: It travels through the internet
3. Finally: It arrives at a big computer in the cloud!
4. Later: When you open your phone app, the cloud sends the info back to you!

106.3.5 Cloud Words for Kids

Word	What It Means
Cloud	Big computers far away you can use through the internet
Data Center	A special building full of computers
Upload	Sending stuff TO the cloud
Download	Getting stuff FROM the cloud
Storage	A place to keep your data

106.3.6 Fun Fact!

When you watch a YouTube video, it’s not stored on your tablet - it comes from the cloud! The video lives on Google’s computers, and they send it to you when you press play!

For Kids: Meet the Sensor Squad!

Cloud computing is like having a super-smart friend with a giant brain who lives far away!

106.3.7 The Sensor Squad Adventure: The Case of Too Many Memories

One day, the Sensor Squad was collecting SO much data that their tiny brains couldn’t remember it all! Sunny the Light Sensor was tracking sunrise to sunset every single second. Thermo the Temperature Sensor was measuring hot and cold 100 times per minute. Motion Mo was detecting every little movement in the whole house!

“Help!” cried Power Pete the Battery Manager. “We’re running out of space to store all these numbers, and I’m getting tired carrying all this data around!”

That’s when Signal Sam the Communication Expert had a brilliant idea. “I know some SUPER powerful computers that live in special buildings far away! They have rooms and rooms full of memory. Let’s send our data there through the internet!”

And that’s exactly what they did! Now whenever the Sensor Squad collects data, Signal Sam sends it zooming through the internet to the CLOUD - giant buildings full of computers that never forget anything. When the family wants to see what temperature it was last Tuesday at 3pm, the cloud remembers! The Sensor Squad can now focus on sensing, while their cloud friends handle all the heavy thinking.

106.3.8 Key Words for Kids

Word	What It Means
Cloud	Super powerful computers in special buildings that store your data through the internet
Data Center	A building full of computers that never sleep - they keep your information safe 24/7
Upload	Sending your information UP to the cloud, like mailing a letter

106.3.9 Try This at Home!

Cloud Memory Game: Close your eyes and try to remember what you had for breakfast every day last week. Hard, right? Now imagine remembering EVERY breakfast for the past 5 years! That’s what the cloud does - it remembers everything so our small devices don’t have to. Ask a parent to show you a photo app on their phone. All those thousands of photos are stored in the cloud, not just on the tiny phone!

106.4 Getting Started (For Beginners)

New to Cloud Computing? Start Here!

This section is designed for beginners. If you’re already familiar with cloud concepts, feel free to skip to the technical sections below.

106.4.1 What is Cloud Computing? (Simple Explanation)

Analogy: Think of cloud computing like renting vs. owning a car.

Approach	Car Analogy	Computing Equivalent
Own everything	Buy a car, garage, tools	Buy servers, build data center
Rent as needed	Uber/Lyft when you need a ride	Use cloud when you need compute

The Cloud = Someone else’s computers that you rent by the hour

106.4.2 Why Does IoT Need the Cloud?

Your smart home has 50 devices. Where should data go?

Figure 106.1: Comparison of local vs. cloud deployment options for IoT devices.

106.4.3 How IoT Data Flows to Cloud

Figure 106.2: IoT to Cloud Data Flow: Sensor data travels through gateways and the internet to cloud services for storage, analysis, and automated actions.

106.4.4 Self-Check Questions

Before continuing, make sure you understand:

1. What’s the main advantage of cloud for IoT? (Answer: Scalability - handle millions of devices without buying hardware)
2. Why might you NOT use cloud for IoT? (Answer: Latency concerns, data privacy requirements, or unreliable internet)

Key Takeaway

In one sentence: Cloud computing provides virtually unlimited scale and powerful analytics for IoT, but introduces latency and connectivity dependencies that make it unsuitable for real-time control.

Remember this rule: Use cloud for storage, analytics, and management; use edge for real-time decisions and offline operation.

106.5 Introduction

Cloud Computing has become a fundamental enabler for Internet of Things (IoT) systems, providing the scalable infrastructure needed to store, process, and analyze massive volumes of sensor data. The combination of IoT’s distributed sensing capabilities with cloud computing’s centralized processing power creates powerful applications across domains.

This chapter explores cloud computing fundamentals and the NIST model that defines cloud characteristics.

NIST Definition of Cloud Computing

Figure 106.3: Visual model of cloud computing showing on-demand resources, scalability, and service delivery over the internet

“Cloud computing is a model for enabling convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.”

– NIST Special Publication 800-145

Figure 106.4: NIST cloud computing definition mindmap showing three main dimensions.

106.6 Evolution: From Grid to Utility to Cloud

Figure 106.5: Evolution of cloud applications from grid computing through utility computing to modern cloud services

Figure 106.6: Comparison of grid computing, utility computing, and cloud computing characteristics

Figure 106.7: Cloud computing evolution timeline.

Key Differences:

Aspect	Grid Computing	Utility Computing	Cloud Computing
Focus	Scientific HPC	Metered resources	On-demand services
Granularity	Coarse-grained jobs	Resource-level	Service-level
Access	Limited community	Metered users	Public/private/hybrid
Abstraction	Low (hardware-aware)	Medium	High (infrastructure-hidden)
Elasticity	Limited	Moderate	High

106.7 NIST Cloud Computing Model

The NIST model defines cloud computing through three dimensions: Essential Characteristics, Service Models, and Deployment Models.

Figure 106.8: NIST cloud computing model architecture showing three dimensions.

106.7.1 Essential Characteristics

1. On-Demand Self-Service

• Users provision resources automatically without human intervention
• No need to contact provider for each resource request
• Self-service portals and APIs

2. Broad Network Access

• Resources available over network
• Accessible via standard protocols (HTTP, MQTT, WebSockets)
• Support heterogeneous devices (mobile, desktop, IoT)

3. Resource Pooling

• Multi-tenant model shares resources
• Location-independent resource assignment
• Dynamic allocation based on demand

4. Rapid Elasticity

• Resources scale up/down automatically
• Appears unlimited to consumers
• Responds to traffic spikes instantaneously

5. Measured Service

• Resource usage monitored and metered
• Pay-per-use billing model
• Transparency for provider and consumer

106.8 Knowledge Check: Rapid Elasticity in Practice

Quiz: NIST Definition of Cloud Computing

Scenario: E-Commerce IoT Smart Home Product Launch

You’re the cloud architect for a company launching a new smart home security camera that integrates with a cloud-based monitoring platform. Your capacity planning shows:

Normal Operations (364 days/year):

• Active devices: 1 million cameras
• Backend servers needed: 10 servers (AWS EC2 m5.large @ $0.096/hour)
• Peak concurrent users: 50,000
• Data ingestion rate: 100 MB/second

Black Friday Product Launch (1 day):

• Active devices: 10 million cameras (aggressive marketing campaign)
• Backend servers needed: 100 servers (10x scale)
• Peak concurrent users: 500,000
• Data ingestion rate: 1 GB/second

Traditional Infrastructure Approach:

Capital expense: 100 servers x $3,000 = $300,000 upfront
Annual maintenance: $30,000 (10% of CAPEX)
Power/cooling: $24,000/year
Data center space: $12,000/year
Staff (3 ops engineers): $180,000/year
Total first year: $546,000
Utilization: 11.5% average

Cloud Approach with Rapid Elasticity:

Normal days (364 days): 10 servers x $2.30/day = $23/day
Black Friday (2 days): 100 servers x $2.30/day = $230/day
Annual cost: $8,832/year
Savings: $546,000 - $8,832 = $537,168 (98.4% cost reduction)

Key Insight:

Rapid elasticity means you can: - Scale up: Add 90 servers in 5 minutes via API call - Scale down: Remove 90 servers 48 hours later, stop paying immediately - Pay only for what you use: No idle capacity costs

106.9 Common Misconceptions

Common Misconceptions About Cloud Computing

Misconception 1: “Cloud is always cheaper than on-premises”

• Reality: Cloud is cheaper for variable workloads and small-scale deployments. For stable, predictable workloads at massive scale, on-premises can be 40-60% cheaper after 3+ years.

Misconception 2: “Cloud means no security responsibility”

• Reality: Shared responsibility model means customers manage application security, access control, and data encryption even in SaaS.

Misconception 3: “All cloud services auto-scale infinitely”

• Reality: Services have default rate limits. You must request limit increases weeks in advance.

Misconception 4: “Multi-cloud avoids vendor lock-in”

• Reality: Multi-cloud adds complexity (2x operational burden, cross-cloud data transfer costs).

Misconception 5: “Edge eliminates the need for cloud”

• Reality: Edge handles real-time processing, cloud handles historical analytics, ML training, global orchestration. They’re complementary.

106.10 Key Concepts

Key Concepts Summary

• Cloud Computing: On-demand delivery of IT resources over the Internet with pay-as-you-go pricing
• Essential Characteristics: On-demand self-service, broad network access, resource pooling, rapid elasticity, measured service
• Data Centers: Facilities housing servers, storage, and networking equipment that provide cloud computing services
• Elastic Scalability: Ability to rapidly scale computational resources up or down based on demand

106.11 Worked Example: On-Premises vs Cloud for an IoT Fleet Management Startup

Scenario: A logistics startup in Manchester tracks 500 delivery vans with GPS + OBD-II sensors. Each van reports location (every 10 sec), engine diagnostics (every 60 sec), and driver behaviour events (acceleration, braking – sporadic). They expect to grow to 5,000 vans within 18 months.

Data Volume Estimation:

Data Type	Per van	500 vans	5,000 vans
GPS (10 sec intervals, 12 hrs/day)	4,320 msgs/day x 80 bytes = 338 KB/day	165 MB/day	1.65 GB/day
OBD-II diagnostics	720 msgs/day x 200 bytes = 141 KB/day	69 MB/day	690 MB/day
Driver events (~40/day)	40 x 150 bytes = 5.9 KB/day	2.9 MB/day	29 MB/day
Daily total	485 KB	237 MB	2.37 GB
Monthly total	14.6 MB	7.1 GB	71 GB
Annual storage (cumulative)	–	85 GB	852 GB

Cost Comparison (3-Year TCO):

Cost Element	On-Premises	Cloud (AWS IoT Core)
Year 0 (setup)
Server hardware (sized for 5,000 vans)	GBP 28,000	GBP 0
Network equipment	GBP 4,500	GBP 0
Data centre space (colocation)	GBP 6,000/yr	GBP 0
Setup labour (2 engineers x 3 months)	GBP 36,000	GBP 8,000 (1 engineer x 1 month)
Annual operating
Power + cooling	GBP 3,600/yr	GBP 0
Sys-admin (0.5 FTE)	GBP 22,000/yr	GBP 0
AWS IoT Core (messaging)	GBP 0	GBP 840/yr (500 vans) → GBP 8,400/yr (5,000)
Compute (EC2/Lambda)	GBP 0	GBP 1,200/yr → GBP 4,800/yr
Storage (S3 + RDS)	GBP 0	GBP 360/yr → GBP 2,400/yr
3-Year Total	GBP 163,300	GBP 42,680

Break-Even Analysis:

• On-premises requires GBP 74,500 upfront (CapEx) before a single van is tracked
• Cloud starts at GBP 8,000 setup + GBP 200/month (OpEx) for 500 vans
• The startup preserves GBP 66,500 in cash for marketing, hiring, and van acquisition
• At 5,000 vans, cloud costs rise to GBP 1,300/month – still 3.8x cheaper than on-premises over 3 years

Putting Numbers to It

When does cloud TCO beat on-premises? Let’s calculate the exact break-even point for the fleet management scenario:

For the 3-year comparison:

• On-premises total cost: \[\text{CapEx} = £74{,}500 \text{ (upfront)}\] \[\text{OpEx} = 3 \times (£6{,}000 + £3{,}600 + £22{,}000) = £94{,}800\] \[\text{Total} = £169{,}300\]

• Cloud total cost (scaling from 500 to 5,000 vans): \[\text{Year 1} = £8{,}000 + 12 \times £200 = £10{,}400\] \[\text{Year 2} = 12 \times £650 = £7{,}800\] \[\text{Year 3} = 12 \times £1{,}300 = £15{,}600\] \[\text{Total} = £33{,}800\]

Savings: \(£169{,}300 - £33{,}800 = £135{,}500\) (80% cheaper). The cloud’s elastic scaling – paying \(\$200/\text{month}\) initially instead of \(£74{,}500\) upfront – preserves £66,500 in startup capital for growth investments.

Key Insight: Cloud computing’s “rapid elasticity” (NIST Characteristic #4) is transformative for startups. The fleet management company pays for 500 vans today and scales to 5,000 without purchasing new hardware. The on-premises approach requires sizing for peak capacity on day one – wasting 90% of capacity for the first 18 months.

Decision Framework: When Does On-Premises Become Cheaper?

Question: At what scale does on-premises TCO become lower than cloud?

Break-Even Formula:

Cloud monthly cost: C_cloud = (devices × messages_per_day × 30 × price_per_million) / 1,000,000
On-prem monthly cost: C_on_prem = (hardware_depreciation + power + admin) / 36

Break-even when: C_cloud > C_on_prem

Calculated for this scenario:

Device Count	Cloud Cost/Month	On-Prem Cost/Month	Winner
500	GBP 200	GBP 4,536	Cloud (23x cheaper)
5,000	GBP 1,300	GBP 4,536	Cloud (3.5x cheaper)
50,000	GBP 13,000	GBP 9,072	Cloud (1.4x cheaper)
100,000	GBP 26,000	GBP 15,000	On-Prem (1.7x cheaper)
500,000	GBP 130,000	GBP 48,000	On-Prem (2.7x cheaper)

Key Thresholds:

1. Under 50,000 devices: Cloud is significantly cheaper (3-23x)
2. 50,000-100,000 devices: Cloud and on-prem costs converge
3. Above 100,000 devices: On-prem becomes cheaper IF you have stable workload and in-house expertise
4. Critical factor: On-prem only wins if workload is predictable and stable for 3+ years

Decision Rule: Choose cloud for variable or growing workloads under 100,000 devices. Consider on-premises only when you have: (1) stable device count above 100,000, (2) predictable traffic patterns, (3) in-house infrastructure team, and (4) 3-year commitment to hardware investment.

Common Mistake: Ignoring Hidden On-Premises Costs

The Mistake: A company calculates that cloud messaging costs GBP 10,000/month for 50,000 devices, while buying 10 servers costs GBP 30,000 one-time. They conclude on-premises is cheaper (GBP 30,000 vs GBP 120,000 annual cloud cost) and build their own data center. 18 months later, they’ve spent GBP 180,000 and the system still can’t handle peak loads.

Hidden costs they missed:

Cost Category	Annual Amount	Why Missed
DevOps/SysAdmin salaries (2 FTEs)	GBP 110,000	“Existing staff can handle it”
Backup systems (redundancy)	GBP 30,000	“We’ll add it later if needed”
Network bandwidth upgrades	GBP 12,000	“Current internet is sufficient”
Security patches and monitoring	GBP 8,000	“Open source tools are free”
Unplanned downtime recovery	GBP 20,000	“We won’t have outages”
Total hidden costs	GBP 180,000/year

Actual 3-year TCO: GBP 30,000 (servers) + GBP 540,000 (hidden costs) = GBP 570,000 vs Cloud: GBP 360,000. On-premises ended up 58% MORE expensive.

How to Avoid: Calculate Total Cost of Ownership (TCO) including: (1) Hardware + depreciation, (2) Facilities (power, cooling, space), (3) Network (bandwidth, redundancy), (4) Staff (hiring, training, 24/7 coverage), (5) Security (tools, audits), (6) Downtime costs (revenue impact), (7) Opportunity cost (what else could engineers build?). Rule of thumb: multiply hardware costs by 5-7x to get realistic 3-year TCO for on-premises IoT infrastructure.

106.12 Concept Relationships

Current Concept	Builds On	Enables	Contrasts With	Common Confusion
Cloud Computing	Grid/utility computing, NIST definition	Scalable IoT analytics, elastic storage	On-premises infrastructure	Cloud always cheaper (false – depends on scale)
Rapid Elasticity	On-demand self-service, resource pooling	Auto-scaling IoT backends, burst handling	Fixed capacity planning	Elasticity = infinite resources (still has limits)
Measured Service	Pay-per-use billing, metered resources	Cost optimization, usage transparency	Flat-rate pricing	Metered = expensive (can be cheaper at variable load)
Edge vs Cloud	Cloud latency, bandwidth limits	Hybrid architectures, fog computing	Pure cloud architectures	Edge replaces cloud (they’re complementary)
Break-even Analysis	Cost modeling, TCO calculation	Platform selection, budget planning	Ignoring hidden costs	Hardware cost = total cost (ignores OpEx)

106.13 See Also

• Cloud Service Models - IaaS, PaaS, SaaS models for IoT
• Cloud Deployment Models - Public, private, hybrid cloud strategies
• Cloud Security for IoT - Shared responsibility, IAM, per-device certificates
• Edge-Fog-Cloud Overview - Computing continuum for IoT
• Production Cloud Deployment - Cost optimization, throttling, OTA updates

106.14 Cloud vs On-Premises Break-Even Calculator

Use this calculator to find at what device scale cloud computing becomes more expensive than on-premises infrastructure.

viewof monthlyCloudPerDevice = Inputs.range([0.01, 5.00], {value: 0.35, step: 0.05, label: "Cloud cost per device per month (USD)"})

viewof onpremCapex = Inputs.range([5000, 500000], {value: 45000, step: 5000, label: "On-premises hardware capital cost (USD)"})

viewof onpremOpexMonthly = Inputs.range([500, 20000], {value: 2500, step: 500, label: "On-premises monthly operational cost (USD, incl. staff)"})

{
  // Cloud cost grows with device count; on-prem is roughly fixed (for a given hardware capacity)
  const cloudAt = (n) => n * monthlyCloudPerDevice;
  const onpremAt = (n) => onpremOpexMonthly + onpremCapex / 36; // amortize over 3 years
  
  const breakEven = Math.round((onpremCapex / 36 + onpremOpexMonthly) / monthlyCloudPerDevice);
  const fiveYrCloud_1k = 1000 * monthlyCloudPerDevice * 60;
  const fiveYrOnprem = onpremCapex + onpremOpexMonthly * 60;

  const devicePoints = [100, 500, 1000, 5000, 10000, 50000, 100000];
  const rows = devicePoints.map(n => {
    const c = cloudAt(n);
    const o = onpremAt(n);
    const winner = c < o ? "Cloud" : "On-Prem";
    const color = c < o ? "#16A085" : "#E67E22";
    return `<tr><td style="padding:4px;">${n.toLocaleString()}</td><td style="padding:4px;">$${c.toLocaleString(undefined,{maximumFractionDigits:0})}/mo</td><td style="padding:4px;">$${o.toLocaleString(undefined,{maximumFractionDigits:0})}/mo</td><td style="padding:4px;color:${color};font-weight:bold;">${winner}</td></tr>`;
  }).join('');

  return html`<div style="background: var(--bs-light, #f8f9fa); padding: 1rem; border-radius: 8px; border-left: 4px solid #E67E22; margin-top: 0.5rem;">
    <p><strong>Break-even device count:</strong> ~${breakEven.toLocaleString()} devices (above this, on-premises may be cheaper per month)</p>
    <table style="width:100%;border-collapse:collapse;font-size:0.85em;margin-top:0.5rem;">
      <thead><tr style="background:#2C3E50;color:white;"><th style="padding:4px;">Devices</th><th style="padding:4px;">Cloud/month</th><th style="padding:4px;">On-Prem/month</th><th style="padding:4px;">Winner</th></tr></thead>
      <tbody>${rows}</tbody>
    </table>
    <p style="margin-top:0.5rem;font-size:0.8em;color:#7F8C8D;">On-premises cost is fixed at $${onpremAt(0).toLocaleString(undefined,{maximumFractionDigits:0})}/month (hardware amortization + operations). Cloud scales linearly with device count.</p>
  </div>`;
}

Interactive Quiz: Match Concepts

Interactive Quiz: Sequence the Steps

Common Pitfalls

1. Treating Cloud as Infinitely Fast

Assuming cloud services have zero latency because they are “just a computer in the sky.” Round-trip latency to a cloud region is typically 20–100 ms, making cloud unsuitable for real-time IoT control loops requiring <10 ms response. Use edge computing for time-critical decisions.

2. Ignoring Service Quotas Until Production

AWS IoT Core defaults limit connections to 500,000 devices per account. A team designs for 1 million devices without requesting quota increases. At launch, device connections are throttled and data is lost. Always check and request service limit increases 4–8 weeks before production launch.

3. Misunderstanding the Shared Responsibility Model

Believing the cloud provider secures everything. AWS secures the physical infrastructure and hypervisor; the customer is responsible for IAM policies, data encryption, network security groups, and application-level security. IoT device credentials and access control are always the customer’s responsibility.

4. Optimizing for Peak Load Instead of Average Load

Provisioning cloud resources for maximum possible concurrent IoT device connections. At average load this wastes 80–90% of spend. Use auto-scaling groups and serverless (AWS Lambda, Azure Functions) that scale to zero when idle and scale up automatically for bursts.

Label the Diagram

💻 Code Challenge

106.15 Summary

This chapter introduced cloud computing fundamentals:

1. Definition: Cloud computing provides on-demand, scalable IT resources over the internet
2. Evolution: Grid computing -> Utility computing -> Modern cloud
3. NIST Model: Five essential characteristics define true cloud computing
4. IoT Relevance: Cloud enables storage, processing, and analytics for massive IoT data volumes

106.16 Knowledge Check

Quiz: Cloud Computing for IoT

106.17 What’s Next?

Now that you understand cloud computing fundamentals, continue with:

Next Topic	Description
Cloud Service Models	IaaS, PaaS, and SaaS for IoT applications
Cloud Deployment Models	Public, private, and hybrid cloud strategies