The characteristics of public clouds

As with any computing service, cloud services run on physical servers, storage, and network hardware in a data centre. The hyper-scale cloud service providers operate their compute resources from groups of vast, dedicated data centres at strategically selected locations around the world. These locations are carefully selected based on a number of criteria, including proximity to renewable energy sources and population centres and the level of geographical and geopolitical stability.

From the perspective of the consumer, the complexity of operating this vast global infrastructure of data centres, servers, and services generally referred to as their cloud platform is the responsibility of the cloud provider. Cloud platforms remove much of the complexity of running services and workloads for the consumer, leaving them to focus on the capabilities that add value to their organisation. The consumer does not need to understand how the cloud service is run, but simply how to use it.

Consumers interact with the cloud platform through a web-based management portal, or APIs. This enables you to provide services and infrastructure (such as virtual servers or databases), view billing information, configure billing and reporting, etc, in an automated and agile manner.

It is also useful to understand the characteristics of cloud services.

• on-demand self-service: you can self-provision computing capabilities without interaction with the provider. These activities are completed remotely over the internet using a web browser, command-line interface or APIs
• broad network access: the services are available over the internet, or through a direct private connection
• rapid elasticity: you can quickly scale and delete provisions without the need for the time consuming purchase of hardware, or the provision of new virtual machines
• measured service: cloud systems can automatically control and optimise resource use by leveraging a metering capability. Resource usage can be monitored, controlled, and reported to provide transparency of service utilisation
• resource pooling: The provider’s resources are pooled to serve multiple consumers using a multi-tenant model by default, although sole-tenant configurations are possible

One of the critical differences in building and managing in public cloud is the concept of shared responsibility. No longer is the organisation and product teams responsible for the end-to-end delivery of technology now the provider takes on a number of responsibilities.

While the split of what exactly is the responsibility of the customer varies between providers and individual services it is documented by the providers. Understanding where their responsibility ends is critical for building and maintaining products in public cloud.

The diagram below shows how AWS defines shared responsibility. In general, the provider is responsible for providing the foundations and providing elements that can be connected together. The customer’s responsibility is in general to understand and use these capabilities to make their products, secure, easy to maintain, scalable, cost effective and reliable.