Over the past decade or so, the public cloud  has fundamentally reshaped virtually every aspect of enterprise operations. But, while plenty have benefited from this shift, perhaps none have enjoyed the convenience of cloud more than the enterprise “end user” — the software developers, data scientists, and other subject matter experts whose work has laid the foundation for much of our modern way of life.

Rather than relying on an overworked IT department to provision the infrastructure, tools, and other resources they need to do their jobs, these high-value, in-demand professionals are now able to spin up entire development environments at-will, with the click of a button. And for a workforce whose skills are both costly and scarce, this liberation from bottlenecks, dependencies, and inefficiencies has been nothing short of transformative. 

Cloud Costs and Compliance Challenges Drive Buzz Around Repatriation

However, the push to the cloud hasn’t come without its challenges. While the “digital transformation” imperative continues to drive business decision makers cloud-ward,  the cost of cloud computing services has been climbing dramatically in recent years — putting a significant pinch on profit margins (especially for more mature businesses, who’ve already scaled and for whom growth has slowed). Analyst house Canalys reports that worldwide cloud infrastructure expenditures saw a year-over-year increase of 28% in Q3 of 2022, and have projected that costs will likely balloon even more dramatically in 2023.

 In addition to rapidly increasing costs, the public cloud also presents a great deal of complexity from a compliance perspective — especially as businesses expand their footprint globally. Together, these forces have led to a steadily increasing buzz around the idea of “cloud repatriation” — the movement of workloads from public cloud infrastructure to on-prem, private, or collocated infrastructure solutions.

Overwhelmed by ballooning cloud costs, the push for repatriation promises to reduce overhead and streamline compliance for businesses that reinvest in on-premises infrastructure. Indeed, a recent survey from Dell revealed that 96% of IT decision makers who repatriated workloads or applications cited cost efficiency as the principal benefit, while an additional 40% listed security and compliance as a primary motivator. 

Hybrid Cloud Strategies and Optimal Workload Placement Prevail

But for all the “ra-ra-ra” around repatriation coming from the server manufacturers of the world, the repatriation phenomenon is less about abandoning the public cloud and more about rebalancing existing hybrid cloud strategies in a way that is advantageous at scale. 

Indeed, according to Cisco’s 2022 Global Hybrid Cloud Trends Report, 82% of organizations have adopted hybrid cloud strategies, while 42% use two to three public IaaS clouds — leaving little doubt as to the dominance of hybrid strategies both now and for the foreseeable future. 

With this in mind, the idea of cloud repatriation as a permanent, one-way shift away from public cloud services is somewhat of a false phenomenon. In reality, what the industry wants and needs more than anything right now is “optimal workload placement” — or, the idea that workloads reside wherever they are best suited, sometimes referred to as the workload’s “best execution venue”. 

In some cases, public cloud services will continue to make sense. In others, however, establishing operations on-premises or in private cloud environments will be the better option. In almost all cases, the best rule of thumb is to bring resources to the data, rather than the other way around. 

The Platform Engineering Imperative

Regardless of the locale or environment, your high-priced developers, data scientists and other subject-matter experts are not at all eager to go back to the “dark ages” of submitting tickets to IT for all of their infrastructure and tooling needs. 

And this is where the promise of platform engineering comes in. For those not already familiar, platform engineering is an approach that aims to make developers’, data scientists’, and other end-users’ lives easier by delivering self-service capabilities with automated infrastructure operations. 

What does that look like in practice? Honeycomb CTO and platform engineering evangelist, Charity Majors, sums up the idea nicely, explaining that one should first “pick a suite of compute and storage options that serve the team’s needs,” and second, “write some tooling that pulls everything into a coherent whole,” which enables end-users to perform all their most essential tasks independently.

By enabling end-users to make the most of their environments — whatever they may be — platform engineering promises to empower organizations to do more with their existing resources, all while retaining top talent and accelerating time to value. Unsurprisingly, organizations are beginning to catch on. According to Gartner, “by 2026, 80% of software engineering organizations will establish platform teams as internal providers of reusable services, components and tools for application delivery”. 

A Deeper Look at Platform Engineering 

So what exactly does a platform engineering team do? The answer to that question will likely vary depending on who you ask. But in my opinion, there are three main pillars that constitute a successful platform engineering program — self-service, platform operations, and platform as a product. 

The first pillar, self-service, is largely self-explanatory. The advent of on-demand cloud infrastructure has made self-service something of an expectation among developers. However, when it comes to internal tools, many developers are still stuck filling tickets and relying on internal IT teams to provide approval and run scripts for them. Platform engineering seeks to change all that, making true self-service a reality regardless of where and what is needed.

The second pillar, platform operations, revolves around the idea of applying a site reliability engineering (SRE) approach to the internal platforms involved. Regardless of where these platforms run, the aim is to operate them as a platform-level service and implement service-level-objectives-style monitoring. 

Finally, the platform as a product pillar refers to bringing a product-management approach to one’s internal platforms. This transforms your developers into “internal customers”, whose needs can then serve as guides for your platform engineering team and reshape one’s IT operations from a “push” model (in which tools, solutions, and other decisions are foisted on developers from the top down) to a “pull” model, in which the development team plays a critical role in these decision-making processes. 

Closing Thoughts… 

It should be clear by now the many and significant benefits that platform engineering has to offer for developers and other end-users. But, they are far from the only beneficiaries of this new movement. For operators (e.g. SREs, DevOps engineers, or DBAs) platform engineering promises to liberate them from the pain of repetitive, tiresome request fulfillment and operation firefighting. Platform engineering effectively frees up operators so that they can engage in more productive and proactive projects and initiatives.

At its core, platform engineering promises to solve the central problem of cooperation between end-users and operators. And with the push for repatriation gaining steam, and the industry gearing up for a rebalancing of hybrid cloud strategies, it’s important that this shift doesn’t come at the expense of efficiency and innovation (or developers’ sanity).