What is cloud computing? Cloud computing is the delivery of computing resources—including storage, processing power, software and databases—over a network, usually the Internet, so people and organisations can use them without managing every underlying machine themselves. It makes computing capacity available on demand, often with usage-based pricing. The cloud is not a place in the sky; it is infrastructure in real data centres operated and maintained by people.
The easiest way to understand cloud computing is to think about a service you use without owning the entire factory that provides it. You can open a web-based email application or store a file in an online account while the service provider operates the servers, storage, networks and much of the software. The practical question is not simply ‘Is this in the cloud?’ but who manages which parts, who can access the data, what happens when the service is unavailable, and what you pay for.
Quick notes: Cloud computing offers shared, configurable computing resources on demand. Its familiar service models are SaaS (use an application), PaaS (develop and run applications on a managed platform) and IaaS (rent infrastructure such as virtual servers). Public, private and hybrid deployments differ. Cloud services can simplify scaling and collaboration, but do not automatically guarantee privacy, backups, low costs or continuous availability.
What cloud computing means: the definition to remember
In the National Institute of Standards and Technology (NIST) model, cloud computing provides convenient, on-demand access to a shared pool of configurable computing resources that can be provisioned and released quickly. The definition includes five essential characteristics: on-demand self-service, broad network access, resource pooling, rapid elasticity and measured service. The formal model gives a clearer test than an advertisement claiming that a service is ‘cloud-powered’.
A home computer can run programs and save files locally. A school can own physical servers in its building. A cloud user can instead request capacity from a provider’s pool. These arrangements can overlap: an organisation may keep sensitive systems on its own infrastructure while using cloud services for other applications. Cloud computing is therefore a delivery and management model, not a synonym for all computing or all Internet usage.
The word ‘cloud’ is a shorthand for infrastructure whose internal complexity is largely hidden from the user. Those servers still require electricity, cooling, physical security, maintenance, networking, software engineering and human expertise. Understanding that physical foundation helps prevent unrealistic expectations.
How cloud computing works, step by step
- A user or application needs a resource. That might be a document, a database, a website, a data-analysis job or extra computing capacity.
- The request reaches a service. A web browser, mobile app or developer’s program connects to an authorised cloud interface over a network.
- The provider assigns resources. Software coordinates computing, storage and networking capacity from one or more physical systems.
- The operation runs. The provider’s infrastructure stores the file, executes the application or processes the requested data.
- A result returns. The user sees the document or answer, while the provider and the customer share different operational responsibilities.
- Usage is measured. Depending on the product, charges may depend on subscriptions, storage volume, compute time, data transfer or other terms.
Not every cloud service uses every one of these steps in the same way. A fixed-price email subscription has a different billing pattern from a virtual machine rented by the hour. A content-delivery network, a managed database and a backup service have different controls. The general model is about accessible services and abstracted resources; the specific contract comes from the provider’s documentation.
The five characteristics of cloud computing
1. On-demand self-service
Users can request a supported service or resource when they need it without negotiating a separate manual allocation each time. For example, a developer may create a test server from a dashboard. Self-service does not mean unlimited access: accounts, quotas, permissions and payment conditions still apply.
2. Broad network access
Services are available over a network using standard mechanisms that can support devices such as computers, phones and tablets. A network connection is still required for many operations, so weak Internet access can reduce the usefulness of an otherwise excellent service.
3. Resource pooling
A provider serves multiple customers using a coordinated pool of resources. Isolation and access controls are intended to keep customers’ workloads and data separate. Sharing physical infrastructure is not the same as giving customers permission to access one another’s information.
4. Rapid elasticity
Capacity can increase or decrease in response to demand. A school registration website may need more capacity during a busy application period than on a quiet day. The ability to scale can be valuable, but a design must still handle bottlenecks, quotas, database limitations and cost.
5. Measured service
The system monitors and reports some resource usage. This can support transparency and usage-related charges. It also means a poorly configured application might consume more billable resources than expected. Understanding the units of measurement is part of using cloud services responsibly.
SaaS, PaaS and IaaS: the difference most beginners need
SaaS — Software as a Service
With SaaS, you use a complete application supplied by a provider. Examples include web-based email, an online document editor or an authorised learning platform. You usually manage your account, content and access decisions; the provider typically manages the underlying application and infrastructure. Exactly who is responsible for security and data protection depends on the product contract.
Question to ask: Can I access my documents, export my work, protect my account and continue if the provider changes its terms?
PaaS — Platform as a Service
With PaaS, developers bring their application code to a provider-managed platform. The provider often takes care of substantial infrastructure and runtime management, leaving the development team to concentrate on application logic, data, permissions and deployment choices. Different platforms include different databases, scaling rules and operating responsibilities.
Question to ask: Which parts of the platform are updated by the provider, and which parts must our team test, secure or maintain?
IaaS — Infrastructure as a Service
With IaaS, an organisation rents infrastructure components such as virtual machines, networks and storage. This gives more control over operating systems and installed software, but also brings more responsibility for configuration, patches, access control, monitoring and recovery. It is often closest to the familiar idea of managing one’s own servers.
Question to ask: Who must patch the operating system, configure the firewall, back up data and restore service after a failure?
Cloud service models at a glance
- SaaS: ‘I want to use an application.’ You primarily manage how you use the service and the information you put in it.
- PaaS: ‘I want to deploy an application without managing all the servers.’ You primarily manage the application and its data, within the provider’s platform rules.
- IaaS: ‘I need configurable computing infrastructure.’ You manage more of the software environment and its security.
- Serverless and managed services: These are additional ways providers package computing and operations. ‘Serverless’ does not mean there are no servers; it means the user typically does not administer the individual servers.
These categories are useful teaching tools rather than perfect boxes. Real cloud products combine responsibilities, and terms can differ across providers. Good engineering starts by reading the actual service agreement rather than trusting the category label alone.
Public cloud, private cloud and hybrid cloud
A public cloud makes provider-operated services available to multiple customers through a service offering. A private cloud uses cloud characteristics in infrastructure dedicated to a particular organisation; it may be operated internally or by a third party. A hybrid cloud links distinct environments so that data or applications can move or work across them under defined controls.
The NIST taxonomy also includes a community cloud arrangement, in which an infrastructure supports a specific community of organisations with shared concerns. In everyday speech, vendors may use the deployment labels differently. The important issue is what is shared, who operates the infrastructure, where it is located and which responsibilities remain with the customer.
Imagine a fictional school that keeps student records in a controlled environment but uses a public cloud service for a general school newsletter. That is a different risk profile from uploading confidential student records into an unapproved shared folder. The suitability of a cloud service depends on data, purpose, contracts and safeguards.
Cloud computing versus cloud storage
Cloud storage is a type of service that stores information on remote infrastructure. Cloud computing is the wider model that may provide applications, databases, processing power, networking and storage together. Saving a photo to an online drive involves cloud storage; running a complex simulation on rented processing capacity is another form of cloud computing.
A useful analogy is a library versus an entire learning centre. Storage provides somewhere to keep and retrieve materials; the broader service environment may also provide workspaces, computing tools and procedures. But an analogy should not hide operational responsibilities: online copies need authentication, sharing controls, retention policies and recovery plans.
A worked example: a class uses a shared document
- The teacher creates a document. A cloud application creates or retrieves its working copy from remote infrastructure.
- Students receive permission. Access should follow the school’s authorised sharing policy, not a public link by default.
- Each student edits. The service processes edits and may synchronise a current view across devices.
- The connection fails for one learner. The service may offer offline work or require a reconnection, depending on the application.
- Someone deletes the document accidentally. Recovery depends on version history, retention rules and backups; cloud storage is not itself proof that a recoverable copy exists.
- The class exports a copy. A local file may provide a useful independent record, subject to privacy and school policies.
This example shows the real benefit of cloud collaboration while making the limitations visible. The first task is education: students should know how to find their work, check their permissions and recover from mistakes. Technical complexity is helpful only when it supports that goal.
What happens inside a cloud data centre?
Cloud systems rely on physical processors, storage devices, network links and software. Virtualisation and resource-management technologies can let a physical machine support multiple isolated computing environments. A provider may distribute workloads across servers and locations, monitor failures and replace equipment without the customer directly handling every operation.
Some applications use containers; some use virtual machines; some run through services that hide those implementation details. A beginner does not need to master every architecture to understand the basic idea: the provider converts a large pool of physical resources into services people can request and operate under defined rules.
The presence of multiple data centres can improve resilience, but it is not an automatic guarantee. A regional outage, common software failure, network problem, billing suspension or customer configuration mistake can still interrupt service. Real reliability depends on the specific service architecture and tested recovery procedures.
Cloud security: who protects which part?
Many providers describe shared responsibility: the provider secures the facilities and parts of the cloud service it operates; the customer remains responsible for some combination of accounts, permissions, application configuration and data use. The exact division differs between SaaS, PaaS, IaaS and individual services.
- Account access: use unique passwords and appropriate multi-factor authentication.
- Sharing permissions: avoid public access to private student, customer or business information.
- Data classification: know whether documents include personal or sensitive material before uploading.
- Backups and recovery: test whether deleted or corrupted records can be restored in time.
- Provider and location: understand contractual terms and data-location requirements relevant to your organisation.
- Application security: patch and test the parts your team controls; a provider cannot fix every flaw in your own code.
- Exit plan: know whether and how data can be exported if a service changes or closes.
A phishing email aimed at a cloud account is still phishing. The fact that the data sits on a professionally managed server does not stop a person from approving a deceptive login request. See What is Phishing? for a practical verification and recovery guide.
Cloud costs: why ‘pay as you go’ can surprise people
Cloud services may replace part of the upfront cost of servers with ongoing fees for computing, storage, network transfer or subscriptions. This can reduce the need to buy equipment before demand is known. But the cheapest arrangement depends on how much work is done, how steady the workload is and how services are configured.
Imagine a fictional application that uses a virtual server at $0.10 per hour. At 24 hours per day for 30 days, the compute charge alone is $72, before storage, network transfer, taxes or other service charges. That is a teaching calculation, not a current product price. An application left running unnecessarily may create costs even when no one uses it.
A practical organisation should set budgets and alerts, check billing units, review unused services and avoid assuming that ‘automatic scaling’ always means ‘automatic savings’. For some predictable workloads, a local system or longer-term pricing arrangement may be more suitable.
What is a cloud API?
An API is a documented software interface that allows one program to request a service from another. Cloud services often expose APIs for creating virtual resources, reading data or operating an application. APIs make automation possible, but a valid API request still needs the right permissions and must handle possible errors.
For the complete beginner explanation, read What is an API | How Software Communicates and Shares Data. A student can understand both concepts by remembering: the cloud is one way computing resources are provided; an API is one way software requests operations.
Cloud computing, artificial intelligence and the Internet
Cloud providers may supply powerful processors, data storage and tools used to train or operate AI systems. But cloud computing is not the same as artificial intelligence. A spreadsheet stored online can use cloud services without involving AI. An AI model may also run locally on a user’s device without a network connection.
Similarly, the Internet and cloud computing are related but not identical. The Internet is an interconnected network of networks. Cloud computing is a service model that can run over the Internet or another suitable network. Calling every connected computer ‘the cloud’ blurs the difference between infrastructure, application and delivery model.
Can cloud computing work without the Internet?
Cloud services require network connectivity between users or applications and the relevant resources, but that network does not always have to be the public Internet. Some applications offer offline modes that store temporary local copies and synchronise later. Offline support is a feature of the particular app, not a universal property of cloud computing.
If students must work during an unreliable connection, choose tools that support appropriate offline access, save independent copies where permitted, and define how changes will be merged. Planning for failure is more useful than a vague promise that a service works ‘anywhere’.
Cloud computing in school, work and everyday life
For a student
Common uses include course documents, shared writing, digital homework submission, remote software and educational simulations. Students should learn the account, permission and backup rules rather than assume every school file is automatically permanent.
For a small business
A business may use hosted email, inventory software, accounting tools or a website without maintaining its own server room. That flexibility is attractive, but the business still needs to understand costs, account ownership, customer data, availability and how to switch providers.
For a developer
A developer can deploy code, rent computing capacity, connect storage and use managed databases. The challenge is to move beyond the ‘hello world’ demonstration and design for errors, data privacy, scaling, monitoring and safe deployment.
For a researcher
Researchers may process large datasets with temporary computing capacity. They must still check measurement definitions, legal permissions, methodological reproducibility, data quality and whether their work can be repeated under the same conditions.
A quick decision checklist: do you need the cloud?
- Purpose: What exactly must the software accomplish?
- Users: Who needs access, from where and with which devices?
- Data: Is it public, private or subject to specific rules?
- Availability: What happens during an outage?
- Security: Who controls accounts, permissions, updates and backups?
- Cost: What will be billed, and what happens if use increases?
- Portability: Can you export, move and recover the important information?
A cloud solution is a good solution only if it fits the actual task. A simple local tool can sometimes be more economical, private or dependable. Conversely, a managed service may make collaboration and scaling much easier. Technology choice is a comparison, not a contest of fashionable labels.
Common misunderstandings to repair
- ‘The cloud means there are no physical servers.’ Real data centres provide the service.
- ‘Everything in the cloud is automatically backed up.’ Recovery depends on retention, backups, configuration and testing.
- ‘The provider handles all security.’ Users and organisations retain responsibilities for some risks.
- ‘Cloud is always cheaper.’ Total costs vary with workload and service design.
- ‘Cloud storage and cloud computing mean the same thing.’ Storage is one part of the broader model.
- ‘If an app works online, it must be cloud computing.’ Ordinary networked applications do not all meet the cloud model’s characteristics.
- ‘One successful login means all information is safe.’ Permissions, sharing links and phishing still matter.
A seven-day cloud-computing learning plan
- Day 1: Explain the definition of cloud computing in one sentence.
- Day 2: Compare local storage, cloud storage and cloud applications.
- Day 3: Explain SaaS, PaaS and IaaS with three different examples.
- Day 4: Draw a user, Internet connection, provider and data centre; trace a request.
- Day 5: Check sharing permissions and recovery options in an approved learning service.
- Day 6: Calculate a fictional usage-based bill and identify hidden assumptions.
- Day 7: Choose a sensible solution for a new scenario and defend the decision using purpose, safety, cost and reliability.
The aim is to explain how the service works and what decisions remain with the user, not to memorise a list of vendor brands.
Frequently asked questions about cloud computing
What is cloud computing in one sentence?
Cloud computing provides configurable computing services over a network on demand, often using shared resources operated by a provider.
Is Google Drive an example of cloud computing?
An online file-storage and collaboration service is an example of a cloud-delivered application and storage service. The broader field also includes processing power, databases and developer platforms.
What is the difference between cloud and server?
A server is a computer or software service that responds to requests. A cloud is a service model that organises and delivers computing resources, typically using many servers and management systems.
Is cloud computing safe?
It can support strong security, but security is not automatic. Assess the provider, the specific service, user accounts, permissions, data handling, configurations and recovery procedures.
What does cloud computing cost?
It depends on the product: a subscription, per-user fee, storage charge, compute charge or other meter may apply. Read the service’s current pricing and billing rules.
Can cloud computing help students?
Yes, especially through authorised collaboration and access to learning tools. Students should also understand privacy, permission settings, offline limitations and the importance of independent thinking.
What happens when the cloud goes down?
Availability depends on the service and its recovery design. Users may temporarily lose access to applications or data. A tested business-continuity plan and suitable local exports or backups can reduce some risks.
Does using the cloud mean my data belongs to the provider?
Not necessarily. Ownership, access rights, processing permissions, retention and deletion responsibilities come from laws, contracts and product policies. Read the actual terms rather than guessing from the word ‘cloud’.
Continue learning at eduKateSG
To see a school-level example, read G1 Computing: Cloud Storage, Web Forms and Website Templates. For the interface that lets software use online services, continue with What is an API?. For program creation, see What is Python?. These concepts are best learned as distinct pieces that connect to one another.
Primary references and further reading
The definitions and service taxonomy are grounded in the NIST Definition of Cloud Computing (SP 800-145) and the NIST cloud-computing overview. For accessible examples of service-model responsibilities, see AWS’s IaaS, PaaS and SaaS explainer. Terminology and responsibilities can differ across products; these references are educational starting points, not endorsements.