Hybrid cloud: Five key questions in 2021

Hybrid cloud promises the best of both worlds. It offers the flexibility, scalability and economy of public cloud infrastructure, with the performance, control and security of on-premise IT systems.

In practice it is often a compromise, driven by the need to support different business processes, use cases, regulatory requirements, IT skills and budgets.

The US government’s National Institute of Standards and Technology (NIST) defines hybrid cloud as, “cloud infrastructure [that] is a composition of two or more distinct cloud infrastructures (private, community or public) that remain unique entities, but are bound together by standardised or proprietary technology that enables data and application portability”.

For organisations that use cloud technology, this recognises that few will use a single, monolithic cloud supplier. Instead, they are likely to use variants that include cloud infrastructure and software-as-a-service (SaaS) applications.

Although it is not explicit in the NIST definition, hybrid cloud is likely to include some on-premise technology. Cloud suppliers now directly support on-premise IT with cloud-like features, through technology such as AWS Outposts or Azure Stack.

IBM and NetApp, both big players in the space, explicitly include on-premise IT as part of hybrid cloud. IBM says it “combines and unifies public cloud, private cloud and on-premise infrastructure to create a single, flexible, cost-optimal IT infrastructure”.

NetApp defines hybrid cloud as “a mixed computing, storage and services environment made up of on-premise infrastructure, private cloud services and a public cloud”. However, Grant Caley, NetApp’s chief technology officer for UK and Ireland, narrows it down further, suggesting a true hybrid cloud involves an application that uses on-premise and cloud- based components.

How is hybrid cloud used?

Hybrid cloud is best used to add agility or scale to existing IT processes.

If the entire application can run as well in the cloud or on-premise, there is no need for the added complexity of a hybrid option. Organisations turn to hybrid cloud to add scale and flexibility to existing applications, or as a way to modernise their interfaces without fully porting them to the cloud.

One example is an application where a database runs on-premise, for performance, security and regulatory reasons, but the customer-facing, or “web-scale”, front end runs in the cloud.

Organisations also use hybrid cloud to connect multiple cloud applications and core business systems together. With more firms using SaaS applications such as Office365 or Salesforce alongside existing or legacy on-premise systems, some organisations find they use hybrid cloud by default. 

Hybrid cloud: What are the leading use cases?

Currently, the most significant application for hybrid cloud is business continuity and disaster recovery, where core data in on-premise systems is replicated in the cloud.

This is a relatively easy path for organisations to take. They can close a secondary datacentre and switch its workload to the cloud, but maintain the production system on-premise.

Hybrid cloud also allows organisations that could not afford a traditional, secondary disaster recovery site to create active-passive failover at a much lower cost.

Hybrid cloud is also being used to modernise applications and extend the use of legacy systems.

But chief technology officers point out that there are applications that will never move to the cloud.

This can be because of their architecture or because the business case does not justify it, perhaps because they expected to last for only a few more years or support a small number of users. In these cases, organisations can develop apps or web-based user interfaces in the public cloud that connect to the business’s systems of record, such as an enterprise resource planning system.

Some organisations moved to this model during lockdown, as an efficient alternative to connect employees working from home to on-premise systems via a VPN. This extends to use cases where, for security or regulatory reasons, organisations opt to keep core data in-house, but connect to the cloud for scale. A further hybrid use case is bringing business data into the cloud for analytics, business intelligence or even machine learning and artificial intelligence. 

What are the pitfalls of a hybrid cloud architecture?

First and foremost, the pitfalls come with complexity. Organisations have to manage two sets of IT infrastructure, including potentially multiple cloud providers, and either physical premises or a datacentre or colo supplier for the on-premise component.

Unless the IT team simply treats cloud and on-premise as two separate technology buckets each running separate applications, software will need to be designed to handle the multiple locations.

This means adaptations at the application layer, and at the OS and storage or file system layer too. Each component needs to be aware that workloads and data will move to and from the cloud.

Reliable bandwidth is essential, too. But even with good bandwidth, applications that need low latency will not be best suited to a hybrid approach.

And, although hybrid and multiple cloud options bring benefits in terms of redundancy, chief information officers (CIOs) will be aware this means managing multiple suppliers’ contracts and multiple system management tools.

One way to determine if a workload suits hybrid cloud is to look at so-called data gravity. If an application’s primary data store is on-premise, then an on-premise or hybrid solution will work best.

If data is primarily in the cloud – as may be the case with content and media, e-commerce, and even internet of things applications – then a cloud-first approach could be better. For applications that need tight control over risk or performance, an on-premise or all-cloud option is usually preferred. 

Where next for hybrid cloud?

Hybrid- or multi-cloud approaches are likely to grow in importance, not least because supplier lock-in is as much an issue in cloud as it is in on-premise technology.

This has prompted CIOs and application suppliers to break down software and workloads so they are more easily portable between clouds, and between cloud and on-premise architecture.

This could include moving an entire application stack from one environment to another, to bursting capacity into the cloud for additional performance or to deal with peak demands.

Containers and other microservices architecture support this, as do the cloud suppliers’ on-premise hardware options. The key is to ensure that data is portable and protected – and achieving this is still a work in progress.