Learn
A1.1.9 | CLOUD COMPUTING SERVICES
01 | WHAT ARE YOU ACTUALLY RENTING?
A school wants shared document editing. A student wants to publish a Python web application. A research team wants machines on which it can install a particular operating system and software stack. All three can use cloud services, but they need different levels of control.
Cloud computing provides access over a network to configurable computing resources or services. These can include processing, storage, platforms and complete applications. Resources can be provisioned as needed, subject to the service’s limits and arrangements.
The three service models in this objective are software as a service (SaaS), platform as a service (PaaS) and infrastructure as a service (IaaS). Their main difference is what the provider supplies and which layers the customer must manage.
Start with the customer’s goal: use an application, deploy an application, or configure computing infrastructure. That question is more useful than assuming every cloud product is just online file storage.
02 | A COMPUTER SERVICE HAS SEVERAL LAYERS
A web application needs more than its visible page. It depends on application code, a runtime that executes it, an operating system and underlying computing and networking hardware. Someone must maintain each relevant part.
A runtime is an environment that executes a program, such as a supported Python runtime. A virtual machine is a software-defined computing environment that can run its own guest operating system. These concepts help distinguish PaaS from a common IaaS arrangement.
Cloud service models move the management boundary. In SaaS, the provider operates the application itself. In PaaS, the customer deploys an application onto a managed platform. In IaaS, the customer receives fundamental computing resources and manages more of the software stack.
SaaS
USE THE SOFTWARE
A ready-to-use application.
Configure the permitted settings and manage how your organisation uses it.
PaaS
DEPLOY YOUR APPLICATION
A supported application-hosting platform.
Supply your code and configuration while the provider operates the underlying platform.
IaaS
CONFIGURE YOUR ENVIRONMENT
Computing, storage and networking resources.
Configure the guest operating system and application stack in a typical virtual-machine deployment.
Generally, moving from SaaS towards IaaS gives more control over the software environment and more management responsibility. It does not mean automatic improvement in every situation.
03 | SaaS: SOFTWARE AS A SERVICE
SaaS provides an application operated by the service provider. Users access the application through an appropriate interface, often a browser or client application, without installing and maintaining its server-side software stack themselves.
Examples include a hosted document-editing suite, an online learning application or a hosted customer-management system. The school uses the supplied features rather than building and running the whole service.
The provider maintains the application and its supporting platform. The customer still manages its own account membership, permitted configuration, data use and sharing choices. A school administrator can often assign roles or settings, but cannot normally replace the provider’s server operating system.
SaaS can make a standard task quick to adopt. Its trade-off is dependence on the service’s available features, terms, interfaces and export capabilities. A configuration option is not the same as unrestricted access to the application’s source code or underlying servers.
04 | PaaS: PLATFORM AS A SERVICE
PaaS provides a managed environment for deploying applications. A developer supplies application code and relevant configuration, using the platform’s supported languages, runtimes, services and deployment methods.
For example, a student builds a Flask project portal and deploys it onto a managed Python web platform. The student controls the routes, templates and application behaviour, while the provider maintains the supported hosting platform and underlying operating environment.
This reduces the need to configure and patch a general-purpose server from scratch. The developer must still maintain their application code, its own libraries and dependencies, secrets, data-handling rules and access checks as required by the service.
The platform’s boundaries matter. An unsupported runtime, unusual system library or requirement for unrestricted operating-system changes may not fit a particular PaaS. You gain application control without necessarily gaining administrator access to the host operating system.
05 | IaaS: INFRASTRUCTURE AS A SERVICE
IaaS provides fundamental resources such as computing capacity, storage and networking. A common example is renting virtual machines and virtual storage rather than purchasing and maintaining physical servers.
In that arrangement, the customer chooses or installs a guest operating system, configures server software and deploys applications. The provider operates the physical infrastructure and the service’s underlying virtualisation layer.
This can suit an organisation needing a particular supported operating system configuration, custom server software or a more flexible environment than an application platform permits. The customer does not usually control the provider’s physical racks or hardware.
Greater control brings more work. In a typical virtual-machine service, the customer must maintain the guest OS, patch its software, configure relevant network access and arrange appropriate application operation and data recovery. IaaS is not simply PaaS with every responsibility automatically handled.
06 | COMPARE THE MANAGEMENT BOUNDARY
| Layer or decision | Typical IaaS | Typical PaaS | Typical SaaS |
|---|---|---|---|
| Physical servers and data-centre infrastructure | Provider | Provider | Provider |
| Underlying virtualisation where used | Provider | Provider | Provider |
| Guest or platform operating environment | Customer manages guest OS | Provider manages platform OS | Provider |
| Provider-supported application runtime | Customer manages installed runtime | Provider | Provider |
| Customer application code and its dependencies | Customer | Customer | Provider operates the supplied application |
| Permitted user/account and sharing configuration | Customer | Customer | Customer |
| Decisions about data use and appropriate access | Customer responsibilities remain | Customer responsibilities remain | Customer responsibilities remain |
This table is a teaching summary, not a contract for every product. The boundary varies with the service. For example, a provider-managed runtime does not mean the provider maintains every library a developer includes in their own application.
Data is also not a single all-or-nothing responsibility. The provider may operate storage and protection mechanisms while the customer decides who should be allowed to access a document. Both parts matter.
07 | THE SAME SCHOOL PORTAL, THREE POSSIBLE APPROACHES
Suppose a school needs students to submit projects and teachers to give feedback. With SaaS, it adopts an existing learning application and configures the available submission features. There is little server administration, but the workflow must fit what that application supports.
With PaaS, developers build a custom portal and deploy it on a supported platform. They control the submission and feedback logic, while the provider operates much of the hosting environment. Application maintenance remains their responsibility.
With IaaS, the team runs its own software stack on rented computing resources. It has more control over the guest environment, but also needs the expertise and time to maintain it.
These are alternatives, not compulsory steps through which every organisation must progress. The school could also combine services. What matters is the required control, available expertise and operational requirements.
08 | EXPLORE WHO MANAGES WHAT
Choose a service model and a scenario. The activity updates the typical layer split and explains how that choice fits the requirement. Compare the amount of software you can change with the amount you must maintain.
The scenarios are illustrative rather than product recommendations. Real decisions require checking the particular service’s support, configuration options, quotas and terms.
| Layer or decision | Who manages it in this example? |
|---|
09 | CONTROL AND FLEXIBILITY HAVE A MANAGEMENT COST
Control means what the customer can configure or change. Flexibility means how well the service can adapt to the application’s needs. These are not identical to speed or quality. A flexible virtual-machine environment can still be badly configured or unsuitable for the organisation maintaining it.
SaaS generally gives the least control over the underlying software stack but can offer useful application settings, integration interfaces and convenient administration. PaaS gives control over deployed applications within platform constraints. IaaS gives more guest-environment control in a typical deployment.
A team needing a standard document editor may gain little from maintaining its own OS and server software. A team needing an unusual runtime may find that the convenience of one PaaS comes with a restriction that prevents its application working.
Explain the trade-off in context: more control can satisfy a custom requirement, but responsibility for maintenance, configuration and operation increases. Less control can reduce that burden when the supplied service already fits.
10 | RESOURCE MANAGEMENT AND ELASTICITY
Resource management means deciding what computing capacity, storage or service allowances are needed and arranging how they are used. Cloud services can make provisioning resources easier, but the customer’s available controls differ by model.
In IaaS, a team may select virtual-machine sizes, storage volumes and network arrangements. In PaaS, it may choose application instances or supported scaling settings. In SaaS, it may choose account numbers, storage allowances or service tiers rather than server details.
Scalability is the ability to handle a larger workload by changing resources or design. Elasticity involves adjusting resources as demand changes, including reducing them when less is needed. Neither is a promise that an application scales automatically without configuration.
A school event may cause a brief traffic peak. More application instances might help a supported PaaS deployment, but quotas, startup time, database limits, session storage and cost can constrain the result. A virtual machine or SaaS account is not unlimited capacity.
11 | AVAILABILITY IS MORE THAN HAVING A CLOUD ACCOUNT
Availability concerns whether the required service can be accessed and used when needed. A reliable provider can improve some aspects of operation, but the full route includes the user’s network connection, authentication, the application and any dependent services.
The customer may need to choose suitable redundancy, backups, recovery arrangements or service options. A provider-operated application does not guarantee that a user’s deleted or overwritten data can always be restored under any desired conditions.
A service-level agreement (SLA) describes agreed service commitments and conditions. It is not proof that interruptions are impossible or that every part of a customer’s workflow is included.
Resource availability also depends on allocation limits, capacity and supported regions or features. A request for more resources may be constrained by quotas or by what the service actually offers. Keep promises separate from capabilities and configuration.
12 | COSTS, DEPENDENCIES AND MOVING BETWEEN SERVICES
Cloud services can reduce the need to buy and operate local hardware, but the total cost depends on use, service features and management effort. Providers may charge for subscriptions, provisioned resources, storage, requests or data transfers; pricing arrangements vary.
A system left running, unexpectedly high use or data moved out of a service can create costs. Cost monitoring, resource limits and shutting down unneeded resources are practical parts of management. Do not equate “cloud” with “always cheaper”.
Moving to another provider can be difficult if an application depends on platform-specific APIs, supported runtimes, proprietary data formats or configuration. This dependence is often called vendor lock-in.
The same organisation can use a mixture of service models. A custom PaaS application might use a separate managed storage service, while staff use SaaS for communication. Classify the service being consumed, not the provider’s entire brand as one category.
13 | SERVICE MODEL IS NOT DEPLOYMENT MODEL
SaaS, PaaS and IaaS describe what kind of service is supplied and the management boundary. Public, private and hybrid cloud arrangements describe how cloud infrastructure is deployed and used. Those are different questions.
A private cloud can provide cloud capabilities for a particular organisation, and access can occur over an appropriate private network. Cloud does not always mean that every component is a public website or that every resource is physically off-site.
Similarly, a website being reachable through a browser does not by itself prove its back end is PaaS or SaaS for its operator. A school might consume the finished website as SaaS while its provider uses IaaS behind the scenes.
Always name the perspective: what is this customer receiving, and which parts can this customer manage? That keeps classifications clear when several organisations and services are involved.
14 | DESCRIBE THE SERVICE AND APPLY IT TO THE NEED
| Requirement | Likely approach to consider | Reason and limitation |
|---|---|---|
| Use a supplied standard application | SaaS | Provider operates the application; customer is limited to supported features and settings. |
| Publish custom code on a supported runtime | PaaS | Customer controls its application without maintaining the whole server stack, but must fit platform constraints. |
| Control a guest operating system and server stack | IaaS | More software-environment control, with more customer administration. |
| Respond to changing demand | Check the chosen service’s resource options | Scaling may be supported, but quotas, application design, configuration and costs still matter. |
For describe, say what is supplied, what the customer controls and what the provider manages. Then connect the model to a real need. “PaaS is for programming” is less useful than explaining that developers deploy their application to a provider-managed platform.
A strong scenario answer might be: “PaaS suits the custom portal if it supports the required Python runtime. The team manages its code, dependencies and access rules while the provider maintains the underlying platform, reducing server administration but limiting OS-level customisation.”
Use the Questions tab to practise these distinctions. The next topic begins data representation, including integer representation in binary and hexadecimal.
Further reading: NIST’s cloud service definitions · Shared responsibility examples