Succeeding with hybrid cloud deployments

More than a decade since public cloud platforms exploded into our consciousness with surging year-on-year growth, a curious thing has happened. Somewhere along the way, the loud, repeated rhetoric about a public cloud-only future has subsided into nary a whimper.

To be clear, there is no disputing that the cloud is now squarely in the mainstream. Both startups and enterprises are using the cloud more than ever, fuelled by an insatiable demand for digital services. Yet there is a dawning awareness that some workloads simply work better in a private cloud and some on-premises deployments should stay where they are.

One of the top concerns around the public cloud would probably be the fear of inadvertent lock-in. Integrate an application tightly enough with a public cloud platform, and one might eventually find it too difficult to migrate to another cloud without a rewrite. This fear of supplier lock-in is why some have called for public sector organisations to adopt a multi-cloud approach from the get-go.

Then there is that occasional bill shock from unchecked public cloud use, often followed closely by frenetic cost-cutting. Other concerns are more insidious, such as differing service level agreement (SLA) standards between clouds. At least one public cloud giant does not offer an SLA for deployments on the same availability zone, while what qualifies as an outage might differ across providers.

Complexity, governance, and transparency of spending are some of the key challenges with the public cloud, notes Daphne Chung, a research director at IDC. In response to a query from ComputerWeekly, she observed that organisations are increasingly deploying cloud management tools to address public cloud deployments.

“Security and compliance are some of the issues that often looms large, so having a policy-driven approach that leverages technology to address these is something users need to look towards deploying,” she said. Another big challenge would be the integration and deployment of the cloud across the broader enterprise. Not doing that well can limit the value of the cloud: “Organisations need to plan their journey and develop a strategy and roadmap to achieve the business value and reach enterprise goals.”

For now, the industry has settled into a quiescence acceptance that a hybrid cloud future is the most likely outcome. And with the low-hanging fruits for public cloud deployments already done and dusted, larger transitions hampered by concerns over cloud lock-ins, even the public cloud giants have rolled out various on-site offerings to address deployments that are staying on-premises for the foreseeable future.

Depending on who you ask, the global hybrid cloud market could reach $171bn in the next five years, with a compound annual growth rate of 21.7% from 2018 to 2025. Yet enterprises cannot run their hybrid cloud as another standalone silo. For maximum gain, they must be operated as part of a cohesive deployment with their burgeoning cloud footprint. And this is where things can get challenging.

“Having management systems across your hybrid cloud environment and to be able to observe, optimise, automate and manage resources in a policy-driven manner is important. One would need the necessary IT capabilities that would support and deliver the hybrid cloud services, such as centralised provisioning, workload portability, consolidated management and operations and centralised security are often required,” said Chung.

Tighter integration between on-premises and the cloud don’t materialise out of thin air, however, and must be meticulously set in place. Aside from ensuring that they have the requisite skills to deploy and run a diverse set of systems, organisations must also ensure that their hybrid cloud deployment is properly integrated.

“Integration is another area of complexity in hybrid cloud. Integration of data, network, management and managing the dependencies of workloads across the cloud platforms will be essential in this area as well,” noted Chung, who pointed out that this starts with a policy-driven approach with proper security and governance.

But this integration work can only be part of the solution. Given the complex interdependencies in a typical hybrid cloud, the ability to keep a close eye on core systems across the entire stack is vital. After all, how else can you figure out why the enterprise resource planning (ERP) system is so slow: is it an application bug, the wide area network (WAN), the back-end database, your cloud-based authentication service – or half a dozen other possibilities?

Success calls for the effective monitoring using best of the breed tools to keep an eye on the core building blocks of the hybrid cloud deployment. This ranges from software-as-a-service (SaaS) offerings, managed database management systems, and cloud-based storage, to traditional on-premises and private cloud systems. And this should be set in place at the start, not as an afterthought.

The old business adage is apt here: “If you can’t measure it, you can’t manage it.”