Cloud repatriation describes a shift away from the public cloud in favor of on-premises infrastructure. What are the reasons for such a move? Is it possible the public cloud isn't the be-all, end-all solution it claims to be? Is this trend genuine, or is it all just vapor?

The answers are actually straightforward. Most businesses start in the cloud, but as they grow, they tend to bring data, workloads, and applications back on-premises to avoid the high cost of scaling. Though companies may return entirely to the public cloud as technology evolves, IT leaders have become more skilled at determining the best strategy for each workload, making selective repatriation the most logical approach to their data needs.

The cloud strategy an organization should adopt depends on its specific situation. Those in need of rapid, frequent scaling should continue to leverage the agility and flexibility of the public cloud, while those who demand tight security and operational control should opt for private, on-premises infrastructure. Still, many organizations may ultimately opt for an even more granular, workload-based, approach to cloud repatriation.

A recent 451 Research study revealed that nearly half of all IT professionals had transitioned a workload or application away from the largest public cloud providers (AWS, Microsoft Azure, and Google Cloud Platform), mainly to self-managed, on-premises data centers. In addition, half of all respondents indicated they planned to do so in the next twelve months.

While cloud repatriation is indeed a genuine trend, it's one that speaks more to the benefits of private, on-premises environments (greater organizational and governance control) than it does to the shortcomings of the public cloud itself. That said, the top three public cloud issues revolve around security concerns (including shadow IT — unsanctioned public cloud usage), application life cycle considerations (production versus test/development environments), and data governance/sovereignty compliance requirements. The adoption of any new technology follows an evolution, so it’s likely a remigration of data, applications, and workloads back out to the public cloud will occur.

In the meantime, the fact remains that countless corporate workloads simply aren't suitable for the public cloud. Suppose your company has selected tighter control and better security of an exclusively, on-premises environment (or hybrid cloud approach). In that case, your organization still needs to strategize its cloud exit strategy and formulate a plan to construct an on-premises data center to reduce costs, maximize power, and ensure flexibility.

When preparing an in-house data center to receive former cloud workloads, IT personnel must consider all possible ramifications. Are power management and monitoring equipment adequate for the changeover? Choosing the right Intelligent PDU can reduce installation complexity by simplifying load balancing and by combining C13 and C19 outlets in a single unit. It will also enable greater flexibility when stacking the racks with the right equipment to receive returning data. Along with flexibility, it's also important to keep watch over (and be alerted on) the environmental conditions of the new IT environment, including humidity, temperature, airflow, leaks, vibration, and more. Environmental sensors identify hot spots and keep equipment cool to minimize downtime, all while adding an important element of sustainability.

In summary, today's organizations are not considering the cloud as their permanent workload destination. Rather, they are viewing various cloud services as a temporary, project-dependent solution, using a back and forth method in a task-based manner. In response to this trend, the top public cloud providers are naturally busy developing varying cloud options (e.g., AWS Outposts, Google Anthos, Azure Stack). These are cloud-to-ground initiatives that offer cloud-tethered, private infrastructure solutions deployed in on-premises data centers, while data center notables, like Dell and HPE are now offering an as-a-service, cloud-like experience in an on-premises package.

However, in all this data movement, one constant exists — the need for flexible power and management. While one workload is comfortably in the cloud, it’s typical that other workloads will be introduced to the on-premises data center. Not everything will be switching back and forth to the cloud. So, when a workload returns, power requirements at the rack may dramatically change. It becomes paramount to have a high degree of power flexibility at the ready to avoid delays when ordering. Heat and humidity at the rack level can also change, requiring careful monitoring to prevent an outage. Think ahead and be prepared when data comes back on-prmises for another stay.