If you haven’t had the chance yet, I highly recommend reading “The Big Cloud Exit FAQ” by David Heinemeier Hansson from 37signals. One thing that really stood out to me was his honesty about the reasons behind the company’s decision to move away from the cloud and his responses to the criticism and questions that followed.
This resonated with me as a cloud expert and architect who has also transitioned some systems from the cloud back to on-premises environments in search of more cost-effective platforms for running certain applications or storing large amounts of data.
The truth is that public cloud platforms may not be suitable for all workloads or data sets, which can often find a more cost-effective home on platforms like your own servers in your own data center. This is not a criticism of cloud computing; it’s simply a reality. The cloud can be expensive and may not be economically feasible for certain applications and data types. Additionally, the cost of hardware, including storage and computing resources, has significantly decreased, as highlighted by Hansson and myself in my recent book.
Facts are impartial
One of the most intriguing aspects of Hansson’s article was his statement, “To say this journey was controversial is an understatement.” The decision to move their applications and data back to traditional hardware sparked a lot of debate and challenged the prevailing bias in favor of the cloud.
I’ve faced criticism whenever I’ve suggested alternatives to cloud platforms, whether it’s during speaking engagements at conferences, panel discussions, or even in meetings with other professionals. The bias against non-cloud solutions that I encountered years ago has now shifted to a mindset of “cloud or nothing,” without considering individual requirements.
As an architect, it’s essential for me to remain open to solutions that best serve the business, regardless of any unfounded biases. Many people fail to see beyond the hype of the latest cloud conferences and fail to recognize that blindly migrating to a public cloud may not always result in the most cost-effective use of resources.
For simple applications or data sets that support a specific set of services, such as providing a single tool on demand, the cloud may not be the most suitable option. Even if it technically works, it could end up costing significantly more without providing any additional benefits compared to hosting the workloads and data on-premises.
Complex deployments that involve a variety of services, such as AI, data management, and security, may justify the investment in cloud solutions. It’s important to reiterate: The cloud isn’t always the best solution for every use case, and our ability to recognize that will yield significant benefits.
We can expect to see more case studies similar to Hansson’s in the future. Numerous technology companies that were initially “born in the cloud” have transitioned back to traditional on-premises infrastructure. While some have publicly announced this shift, many have done so quietly. In all cases, it wasn’t because the cloud couldn’t meet their service quality requirements—it could. Rather, it was a matter of considering more cost-effective alternatives.
Discoveries at 37signals
37signals was a major cloud user with a $3.2 million cloud budget for 2022. The company allocated $600,000 to purchase Dell servers, anticipating significant savings over the next five years.
Naturally, there were inquiries and concerns, which Hansson adeptly addressed in the FAQ. He discussed additional costs related to managing on-premises systems, the limitations of optimization in the cloud, and the company’s approach to meeting security requirements.
Hansson also highlighted the challenges of cost reduction with cloud-native applications and emphasized the importance of a skilled team to address security concerns, which 37signals has in place. Compliance with privacy regulations and the General Data Protection Regulation was cited as a reason for European companies to opt for self-owned hardware instead of relying solely on cloud services, though this isn’t a universal requirement.
Prior to the transition, 37signals conducted thorough assessments of reliability, performance, and cost comparisons. Through prudent financial analysis and strategic hardware investments, the company realized immediate returns and positioned itself for future technological advancements. In essence, their new setup is just as effective, more cost-efficient, and offers greater control, ultimately benefiting the business.
The Cloud Isn’t Always the Solution
There’s no one-size-fits-all solution when it comes to choosing a platform for your systems. The specific requirements of your applications will determine the most suitable environment, rather than following a trend. While the cloud may offer significant advantages in many cases, it’s not always the optimal choice.
There’s a prevailing belief that the cloud is invariably cheaper, more reliable, and more secure. While these arguments can be compelling—and I’ve made them myself—the reality is more nuanced. The cloud is currently a hotbed of innovation, and companies that remain on-premises may find themselves at a disadvantage compared to those leveraging cloud services. This alone could be a compelling reason to migrate to or remain in the cloud.
Ultimately, reality is multifaceted, and the often-maligned “it depends” response from consultants is often the most accurate. Each platform requires thoughtful consideration and planning to ensure you’re on the most cost-effective path. That’s the bottom line.
