Hardware Servers
From the Cloud back to On-Premise

David Heinemeier Hansson’s Cloud Exit

Since David Heinemeier Hansson announced their intention to leave the cloud in October, they have been diligently working towards that goal. Initially, they tried using an enterprise Kubernetes provider, but that didn’t work out, so they decided to build their own tools. Recently, they successfully migrated the first small application out of the cloud. Their next objective is to achieve a complete cloud exit by the end of the summer. According to their initial calculations, this move is expected to save them around $7 million in server expenses over the course of five years, without requiring any changes to the size of their operations team.

Let’s delve into the numbers. In 2022, they spent a total of $3.2 million on cloud services. A significant portion of that amount, almost a million dollars, was dedicated to storing 8 petabytes of files in S3, replicated across multiple regions. This leaves them with approximately $2.3 million for other expenses, including app servers, cache servers, database servers, and search servers. Their plan is to bring these costs down to zero in 2023. The migration of the 8 petabytes from S3 is slated for 2024.

In total, after they place a hardware order with Dell, they estimate an annual expenditure of $840,000, covering bandwidth, power, and hardware amortization over five years. This is significantly lower compared to the $2.3 million spent in the cloud. Moreover, the hardware acquisition enables them to access faster hardware, a greater number of cores, substantially cheaper NVMe storage, and room for expansion at a minimal cost (as long as they can fit four racks per data center).

At this point, they strongly advise any mid-sized SaaS business or larger with stable workloads to compare the costs of renting cloud servers against purchasing their own hardware. Failing to do so could be deemed a financial malpractice. They suggest reaching out to Dell and Deft to obtain real-world numbers and make an informed decision based on individual circumstances.

There are several reasons why a company may consider exiting the cloud and transitioning back to an on-premise data center:

  1. Cost Savings: One of the primary motivations for moving away from the cloud is cost. Depending on the scale of operations and resource requirements, managing an on-premise data center can be more cost-effective in the long run compared to cloud services, especially when considering factors like hardware, storage, and bandwidth costs.
  2. Performance and Control: By managing their own infrastructure, companies have more control over their hardware, network, and storage resources. This increased control allows for fine-tuning and optimization of the infrastructure to meet specific performance needs. It also eliminates the potential for performance issues caused by sharing resources with other cloud tenants.
  3. Data Security and Compliance: Some organizations, particularly those dealing with sensitive data or operating in regulated industries, have strict data security and compliance requirements. Maintaining an on-premise data center allows for greater control over data security measures and compliance adherence, reducing potential risks and ensuring data confidentiality.
  4. Customization and Flexibility: On-premise data centers offer greater customization and flexibility compared to cloud services. Companies can tailor their infrastructure to specific application requirements, software stack, or unique business needs. This level of customization may not always be feasible or cost-effective in a cloud environment.
  5. Predictable Costs: Cloud costs can be variable and unpredictable, especially as usage scales up. By moving to an on-premise data center, companies can have more predictable and stable costs, allowing for better long-term budget planning and financial forecasting.
  6. Reduced Dependency: Relying on the cloud means being dependent on the availability and performance of the cloud service provider. By maintaining an on-premise data center, companies reduce their dependency on external providers, minimizing the risk of potential service disruptions or vendor lock-in.
  7. Legacy Applications and Infrastructure: Some companies may have legacy applications or infrastructure that are better suited to on-premise environments due to compatibility issues or complex dependencies. Moving back to an on-premise data center can provide a more seamless and efficient operational environment for such legacy systems.

It’s important to note that the decision to exit the cloud and transition to an on-premise data center should be carefully evaluated, taking into account the specific needs, resources, and long-term objectives of the organization.