Hybrid cloud

What it means

A hybrid cloud combines on-premise servers or a private cloud with a public cloud, with data and applications moving between them as needed. Sensitive data can stay in-house while computing power, backups or new applications come from the cloud.

Companies choose a hybrid model for various reasons: existing investment in their own infrastructure, regulatory requirements on where data are stored, legacy applications that are hard to move, or the need to add capacity flexibly at peak times without buying more servers.

A hybrid model may include on-premise storage in the company's own server room, space in an external data centre and services from one or more public cloud providers. Using several providers is called multicloud.

How it works in practice

A typical use is backup and disaster recovery to the cloud: production systems run in-house and copies of data are kept in the cloud, from where they can be restored quickly after an outage. It is a relatively cheap way to increase resilience.

An accounting or ERP system may run on-premise while e-mail, collaboration and new web applications run in the cloud. Secure connectivity, unified identity and access management and clear rules on where each type of data may be stored are essential.

The costs of a hybrid model must be tracked as a whole: own hardware, energy and administration as well as cloud fees. FinOps practice helps decide which parts are worth moving to the cloud and which should stay in-house.

What to watch out for

A hybrid model is more complex to manage than a purely on-premise or purely cloud solution. It requires people who understand both worlds, and security settings must also cover the connections between environments, which are often a weak point.

Moving data and applications between environments can be expensive, especially cloud data egress fees. The design should consider the risk of vendor lock-in and include a plan for changing environments if a provider's terms worsen.