Cloud: when you build it, and they don’t come - cracking the utilisation challenge

Who wouldn't want to be more like David Heinemeier Hansson??

In October 2022, 37 Signals' CTO?announced?plans to repatriate his company's data and workloads from the cloud and run its own infrastructure in response to growing cloud costs.

A year later, Hansson?says?he's cut his cloud bill by 60 per cent with more reductions in the pipeline. Along the way, he has spent $500,000 on his own servers and infrastructure.

Cost is consistently voted the biggest challenge in the cloud, so when 37 Signals made the headlines for this move, I had many conversations with senior executives who shared similar concerns.?

However, rather than following Hansson's lead, many companies decided to adopt a mixed cloud and on-prem model: 37 Signals had, in effect, validated hybrid as a strategy rather than a transition phase on the way to full cloud.

Change the game

Repatriation isn't new. Yes, 37 Signals was a great watercooler moment that sparked lots of debate, but analyst IDC?reckons?almost everyone repatriates some of their data, at least some of the time. In the past two years, we've gone from 'cloud only' to 'cloud also'.

And who would turn their back on the cloud? The benefits of flexibility, efficiency, scalability, and innovation are hard to duplicate when building your own infrastructure. Zynga underlined this point when it?returned?to AWS.?

Hybrid might be the new strategy, but costs remain a concern.

It's time to shift the focus in this debate. We need to stop thinking about the cloud as a cost centre and instead consider it a profit centre. Nobody moves to the cloud to save money—they do so to improve their business, generate income, and become more agile.

These goals are realised through utilisation. Many moved to the cloud in the first place because their existing, on-prem IT infrastructure was underutilised, making it expensive and inefficient. High utilisation lowers—or at least offsets—costs.

And it's utilisation where things are sticking.

Not aiming for the moon - yet

We're familiar with the oft-cited 'orphaned virtual machines' or 'over-provisioned clusters' as examples of underutilised infrastructure, but it can be found on a far more extensive and more strategic scale. We see organisations that commit considerable sums to infrastructure that aren't seeing a concomitant pace of development, deployment, or consumption.

There are invariably two reasons for this.

First, the wrong applications have often been prioritised in cloud migrations. There has been a trend to focus on the "big" projects - known as moonshots.

Moonshots, however, undermine utilisation in two ways: first, they lack a mass market of users—they are invariably a single application or workload built for a narrow set of uses. Second, they can be technically complex, carry high risk, and require many heavy controls. Therefore, they divert resources from other applications that could better serve the cause of utilisation.

Utilisation is best served by building the "silent majority" of cloud applications that serve a broader constituency of users by addressing a wide range of uses. These applications are also less technically complex, making them more straightforward to build.

Building these applications also helps IT teams accumulate experience developing and deploying cloud-native applications, paving the way for more ambitious applications. By creating the silent majority, you can establish patterns of consumption that help with forecasting, planning, and analysis.

This strategy means taking applications in tranches—starting with the least complex and most common and moving up the stack. We've seen this approach in banking, where consumer account applications were built first on mobile, decoupled from core infrastructure and delivered quickly to the biggest audience. Applications in more risky, complex activities with greater regulatory and compliance demands, such as derivatives, followed.

Don't delegate

The other stumbling block for utilisation is leadership—right from the top. In cloud projects, it is common for CIOs to sponsor a migration and have their organisation deliver the infrastructure project, then stand back and delegate the role of championing its adoption, letting all applications onboard as departments move to the cloud. CIOs must take a crusading role inside their organisation: lead the charge to sniff out and bring on board those less complex applications with the broadest uses and build that critical mass of the silent majority.

Runaway, unwanted or unaccounted cloud costs are a challenge, but it's important not to allow this to negatively impact our relationship with the cloud. The cloud offers excellent benefits and huge advantages: by tackling utilisation systematically, you can achieve accelerated, organic growth, turning the cloud into a profit centre.