As the calendar turns to 2022, businesses worldwide are preparing for another year of economic uncertainty, anxiety and threat. Those that seize the opportunities created by this shifting landscape will be more successful, and their flexibility and adaptability might even determine whether they survive.

Not surprisingly, then, cloud adoption continues to accelerate as businesses seek agility and new sources of competitive advantage. Cloud spending increased markedly in 2021, as the pandemic placed unforeseen, urgent demands on information technology infrastructure. Gartner anticipates it will accelerate still further, with end-user spending on public cloud services reaching $493 billion in 2022, a year-on-year increase of nearly 22%.

Today, nearly everyone makes at least some use of the public cloud – avoiding it entirely is almost unimaginable. You already have cloud infrastructure; what you need is a cloud strategy. For many IT shops, this means applying governance and cost controls retroactively to infrastructure that’s already in production.

As we do every year, Gartner made six predictions about what’s to come in the year ahead, and what engineers and architects need to do about it. Here’s our outlook for 2022:

Cloud teams will optimize for business outcomes, not technical implementation

Cloud services let smart business leaders respond quickly to opportunities — or threats. So don’t get in the way. You cannot let debates over technical minutiae derail the quick transition to cloud services — even if that means making uncomfortable compromises in initial implementation quality.

However, you cannot abdicate your responsibility to keep business functions safe and highly available. Rather, you must implement flexible governance frameworks that can handle different risk profiles. Optimize cost and risk based on the business needs. You should:

• Build a cloud adoption framework: A good framework is reusable, so that you can onboard new services rapidly while implementing lessons learned and avoiding past mistakes.
• Mitigate risks created by suboptimal cloud adoption: Cloud incurs risks of supply, availability, confidentiality, compliance, concentration and overspending. You must build compensating controls for each.
• Develop a strategic cloud operating model: Formal governing bodies such as advisory councils, communities of practice and a cloud center of excellence will help.

Hybrid and multicloud adoption will increase operational complexity and cost

Everyone wants to say they’re multicloud, but true multicloud architectures are relatively rare. Building an infrastructure that truly spans clouds is complex and expensive. Each cloud provider has proprietary characteristics that give it differentiated value. Cloud services are not commodities; your multicloud architecture must capture the unique strengths of each provider. You don’t want to end up with least-common-denominator functionality across clouds. And you can’t expect any one tool to provide seamless governance, monitoring, asset tracking and security across clouds. You need to:

• Prioritize a primary strategic provider: Choose a primary strategic cloud provider — maybe two, if you have two significantly divergent needs — and master it. Use other providers tactically, only when you have business requirements that cannot be met by your strategic provider.
• Define a workload placement policy: Any multicloud architecture needs a placement rubric that determines which workload goes where. Place workloads primarily based on data and integration affinities. Seek a “good enough” technical fit that satisfies your strategic objectives.

Business resilience will be built into the application architecture

New cybersecurity threats (especially ransomware) and the ongoing organizational disruptions of the pandemic shine a spotlight on IT resilience. Historically, operators built disaster recovery capabilities into the infrastructure. But real IT resilience is a feature of the application itself. Here’s where to start:

• Build resilience into cloud-native apps: Protecting stateful data in Kubernetes means capturing an entire application and its dependencies in such a way as to allow an orchestrated restore in an alternate location or platform. You’ll need pervasive automation, a continuous integration and deployment or CI/CD pipeline, and container-aware backup tools.
• Redesign IREs for ransomware: A secure isolated recovery environment functions as a safe place to conduct restore activities. It is a separate environment from production and Dev/Test. It has dedicated systems, an immutable data vault and no network access to production.

Distributed cloud will displace private and hybrid cloud initiatives

You want cloud services, but you may need them in your data center or at the edge. To achieve this, many IT organizations have attempted to build their own private and hybrid clouds. Most never found real success. But today, you can simply buy a hybrid cloud, rather than building it yourself. “Distributed cloud” extends public cloud services to different locations, while ownership, operation, governance, updates and evolution of the services remain the responsibility of the cloud provider. To exploit distributed cloud:

• Evaluate cloud “substations” that are sold by the major cloud providers, including the hardware and software needed to run public cloud services on-premises.
• Evaluate hyperconverged infrastructure, which provide the full infrastructure stack — virtual compute, storage and networking — on-premises, with hybrid cloud management tools on top.
• Rethink connectivity between locations and devices: Distributed cloud and edge models push the limits of classical approaches to network architecture. You need a network where connectivity follows data sources, using an array of multiple access technologies such as WiFi, 5G/LTE, LTE-M, NB-IoT, LoRA or even satellite links. You might even acquire your whole network from a network-as-a-service or managed network services provider.

Containers and serverless will become an infrastructure foundation for application platforms

Containers provide consistent packaging and streamlined management. Kubernetes is already the industry-standard substrate for cloud portability. Its orchestration capabilities allow a container to be instantiated anywhere, making it the ideal code delivery vehicle across public and private clouds. Take full advantage of these technologies in the public cloud:

• Use provider-native container self-service platforms: The easiest way to deploy a container orchestration platform in production is to use a managed container-as-a-service or CaaS platform in the public cloud, which obviates the need for infrastructure and operations teams to build and support their own container orchestration platform.
• Evaluate “serverless” container platforms: These ingest containers and provision the necessary infrastructure resources automatically, on-demand. They go beyond managed Kubernetes services, which still require the user to configure worker nodes for the clusters. They can be an attractive middle ground between CaaS and function platform-as-a-service models.

The crisis-level skills gap will compromise cloud innovation and execution

Although cloud infrastructure is hardly new, the enterprise IT community simply has not acquired cloud skills fast enough to satisfy the growing demand for cloud services. In many IT organizations, the lack of cloud skills has reached crisis levels. This is often the real barrier to cloud adoption.

Worse, many organizations find it is not feasible to hire cloud experts. Gartner analysis shows that cloud management skills rank among the highest in demand. Qualified candidates aren’t on the job market for long, and they command premium salaries.

Therefore, you’ll need to grow cloud skills internally. To start:

• Prioritize Kubernetes and DevOps skills: These have become core skills for managing cloud-native applications. They are highly transferrable, widely applicable, and largely future-proof.
• Build a talent enablement program: Make skills development part of your organizational goals and individual performance reviews. A formal TEP addresses organization-wide skills gaps and creates an environment of positive change.

Paul Delory is a senior director analyst at Gartner Inc., providing deep technical expertise to systems administrators and the vendors that serve them. His research centers on automation and next-generation infrastructure. He wrote this article for SiliconANGLE.

Image: Larisa-K/Pixabay