Skip to content
VJ Classes

One place to learn science, commerce, arts, computers and life skills

When a company says “we use the cloud”, two questions follow. How much of the work do you want to do yourself? And who else shares the machines you use? The first is answered by the service models, the second by the deployment models.

The layers of a computer system

Every system is built in layers. From the bottom up: the data centre (building, power, cooling); the hardware (servers, disks, network devices); a virtualisation layer that divides one machine into many (Lesson 3); the operating system (such as Linux or Windows); the runtime (the software environment in which a program runs, such as a Python or Java runtime); and finally your application and your data.

In each service model the provider looks after the lower layers and you look after the upper ones. The higher the model, the less you manage.

The three service models

Infrastructure as a Service (IaaS) gives you the basic building blocks: virtual servers, storage and networks. You choose the operating system, install software, apply updates and run the application. You get the most control and the most work. Examples: Amazon EC2, Azure Virtual Machines, Google Compute Engine.

Platform as a Service (PaaS) gives you a ready environment for your code. The provider manages the servers, operating system and runtime; you upload your application and manage its settings and data. Examples: AWS Elastic Beanstalk, Azure App Service, Google App Engine.

Software as a Service (SaaS) gives you a finished application in a browser or app. You only use it, and manage your own data and users. Examples: Gmail and Google Docs (Google Workspace), Microsoft 365, Zoom, and Zoho Books, an accounting product from the Chennai-based Indian company Zoho.

This table shows who looks after each layer:

LayerOn-premisesIaaSPaaSSaaS
ApplicationYouYouYouProvider
DataYouYouYouYou
RuntimeYouYouProviderProvider
Operating systemYouYouProviderProvider
Virtualisation and hardwareYouProviderProviderProvider
Data centreYouProviderProviderProvider

You will also meet serverless computing, also called Function as a Service (FaaS). You upload small pieces of code, called functions, which run only when something triggers them, and the provider handles every server. AWS Lambda and Azure Functions are examples. Serverless does not mean there are no servers; it means you do not manage them.

Which service model fits?

Match each need to a model.

  1. A college wants email and shared documents for 5,000 students. SaaS: Google Workspace or Microsoft 365 needs nothing to install or maintain.
  2. A start-up has a web application written in Python and does not want to patch operating systems. PaaS: it supplies only the code.
  3. A company must move an old accounting server, with its special software, unchanged. IaaS: it needs its own operating system and settings. Moving a system as it is, without redesigning it, is called “lift and shift”.

The deployment models

The service model says what you rent. The deployment model says who owns the cloud and who may use it.

  • Public cloud: owned and run by a provider and open to any customer. The hardware is shared, with each customer logically separated. AWS, Azure and Google Cloud are public clouds.
  • Private cloud: used by one organisation only. It may sit in the organisation’s own data centre or be hosted by a provider for it alone. It gives more control and suits strict data rules, but the organisation pays for the hardware and its upkeep. Open-source software such as OpenStack can be used to build one.
  • Hybrid cloud: a private or on-premises environment connected to a public cloud so that the two work together. A bank might keep customer records in its private cloud and run its public website on a public cloud.
  • Community cloud: shared by several organisations with common needs, such as a group of universities. It is the least common of the four.
  • Multi-cloud: using more than one public cloud provider, for example AWS for storage and Google Cloud for data analysis.

Mixing up hybrid and multi-cloud

Hybrid means private plus public. Multi-cloud means two or more public providers. A company can be both at once, but the words are not interchangeable.

MeghRaj and government cloud

The Government of India has a cloud initiative for e-governance called MeghRaj (GI Cloud), run through the IT ministry, MeitY. The ministry also has a process to empanel cloud providers whose services government departments may use. It is a good example of an organisation choosing which clouds it will trust.

Key points

  • IaaS rents infrastructure (you manage the operating system upwards), PaaS rents a platform (you manage the application and data), SaaS rents a finished application.
  • The higher the service model, the less you manage and the less control you have.
  • Public clouds serve many customers, private clouds serve one organisation, and hybrid clouds connect private and public.
  • Multi-cloud means using several public providers; it is different from hybrid.