What is a public cloud? Definition, examples and trade-offs

A public cloud is computing infrastructure that a provider such as AWS, Microsoft Azure or Google Cloud owns and runs in its own data centres and rents to any customer over the internet. You provision servers, storage, databases and software on demand, pay for what you use, and share the provider's hardware with other customers, isolated by software.

Key takeaways

• NIST defines a public cloud as infrastructure provisioned for open use by the general public, located on the provider's premises.
• You get self-service provisioning, elastic scaling and metered, pay-as-you-go pricing instead of buying hardware.
• Security is shared: the provider secures the cloud itself; you secure your data, identities, configuration and applications.
• The main risks are cost drift, data transfer fees, lock-in and compliance questions such as data location under GDPR.
• Many organisations combine shared cloud services with private or on-premises systems in a hybrid model.

What is a public cloud, according to NIST?

The US National Institute of Standards and Technology gives the most widely used definition. In NIST SP 800-145, a public cloud is infrastructure "provisioned for open use by the general public", owned and operated by a business, academic or government organisation, which "exists on the premises of the cloud provider".

The same document lists five characteristics every cloud service should have:

• On-demand self-service: you create resources yourself, without asking the provider.
• Broad network access: services are reached over the network with standard tools.
• Resource pooling: the provider serves many customers from shared hardware.
• Rapid elasticity: capacity scales out and back in, often automatically.
• Measured service: usage is metered, which makes pay-per-use billing possible.

How does a public cloud work?

The provider builds data centres grouped into regions, installs servers, storage and networks, and exposes them through a web console and APIs. Customers sign up, choose a region and start resources in minutes.

Virtualisation and containers let many customers share one physical machine while staying isolated from each other. Billing is per second, hour, gigabyte or request, depending on the service. Most providers also offer discounts for one- or three-year commitments.

Service models: IaaS, PaaS and SaaS

Public vs private vs hybrid cloud

The difference is who uses the infrastructure and where it runs. A public cloud is shared and provider-run; a private cloud serves one organisation; a hybrid cloud connects the two.

For more on the AWS case specifically, see whether AWS is a private or public cloud.

What are the benefits?

• No hardware purchase: capital spend becomes operating spend, and you stop paying for idle capacity if you size well.
• Elasticity: scale for peaks, such as seasonal sales or month-end batches, and scale back afterwards.
• Speed: teams can test a new idea in hours instead of waiting weeks for servers.
• Reach and resilience: multiple regions and availability zones make it easier to build for high availability and disaster recovery.
• Managed services: databases, queues, AI and analytics services remove much routine operations work.

Risks and how to manage them

Security follows a shared model. AWS describes it as the provider handling security "of" the cloud and the customer handling security "in" the cloud (AWS shared responsibility model). Most breaches come from the customer side: weak identities and misconfiguration.

Common use cases

• Websites, e-commerce and customer apps with variable traffic.
• SaaS products that need to scale with their customer base.
• Data analytics and AI workloads that need large, short-lived compute.
• Development and test environments that can be switched off at night.
• Backup and disaster recovery for on-premises systems.
• IoT platforms that collect data from many devices.

Example: elasticity in practice

Branäsgruppen, one of Sweden's largest ski resort groups, does most of its trade between December and April. On VMware on-premises, its websites had to be scaled by hand and slowed down before each scaling event. After Opsio replatformed the estate onto AWS with auto-scaling and managed databases, the Branäsgruppen case reports 40% better average page responsiveness, 40% lower total cost of ownership and 99.9%+ application uptime.

How Opsio helps

Opsio designs, migrates and runs environments on AWS, Azure and Google Cloud from Karlstad and Bangalore. See our cloud consulting, AWS managed services and hybrid cloud consulting.

Frequently asked questions

Which companies provide public cloud services?

The three largest are Amazon Web Services (AWS), Microsoft Azure and Google Cloud. Others include IBM Cloud, Alibaba Cloud and European providers such as OVHcloud and Hetzner.

Is the public cloud secure?

The major providers run audited, certified data centres, so the infrastructure is generally well protected. Your security depends on how you configure identities, networks, encryption and monitoring on your side of the shared responsibility model.

Who owns the data stored with a cloud provider?

You do. The provider stores and processes it under contract, typically as a data processor under GDPR. You choose the region where data is stored and remain responsible for lawful processing.

Is Microsoft 365 a public cloud?

Yes. Microsoft 365 is software as a service delivered from Microsoft's shared cloud infrastructure. You use the software; Microsoft runs the infrastructure.

Is public cloud cheaper than on-premises?

It can be, especially for variable workloads and when you avoid buying hardware. Steady, always-on workloads can cost more if resources are not right-sized or covered by commitment discounts. See how cloud pricing models work.