Why We Left The Cloud

Summary

In this insightful discussion, the team reflects on their transition from cloud services to in-house servers, revealing the challenges and advantages of managing their own hardware. They emphasize the significant cost savings, improved performance, and the flexibility of using open-source technologies like KVM and Docker. While acknowledging the complexities of self-hosting, they argue that many companies may benefit from evaluating their cloud dependency and considering a similar move. The conversation highlights the balance between scalability and operational efficiency, encouraging businesses to assess their unique needs against the often exorbitant costs of cloud services.

Takeaways

• 💰 Significant cost savings of approximately $1.5 million annually can be achieved by moving from cloud services to in-house servers.
• ⚡ Improved performance for legacy applications is realized by utilizing powerful, dedicated hardware, leading to faster user experiences.
• 🔄 Transitioning to in-house infrastructure can be smoother than expected due to prior investments in containerization and application preparation.
• 👥 The company maintained its operations team size during the transition, raising concerns about future scalability and operational challenges.
• 🛡️ Potential risks, such as DDoS attacks, remain a concern when managing own hardware compared to using cloud providers with built-in protections.
• 🛠️ KVM and Docker are utilized for virtualization and containerization, allowing for effective management without the complexity of Kubernetes.
• 📈 The discussion emphasizes the need for careful engineering considerations when operating in-house servers, balancing costs with resource requirements.
• 🏛️ The company is committed to supporting legacy applications for existing users, despite the challenges of maintaining older systems.
• 🔍 Cloud services can incur high costs, often exceeding in-house operational expenses, making it essential to evaluate specific business needs.
• 🌐 The dialogue encourages businesses to assess their infrastructure strategy carefully, weighing the benefits and drawbacks of cloud vs. in-house operations.

Q & A

What was the main reason for leaving the cloud according to the transcript?

The primary reason for leaving the cloud was the realization that containerizing applications made it easier to exit, allowing the company to bring all their services back to in-house hardware.

How many applications were brought back in-house, and what is their significance?

Six applications were brought back in-house, referred to as Heritage services, which are older applications still used by tens of thousands of customers, generating millions in revenue.

What benefits does the company expect from operating their own hardware?

By operating their own hardware, the company expects to spend significantly less on operations and provide a faster experience for users due to the powerful new hardware.

What open-source tools did the company use for their infrastructure?

The company utilized KVM for virtualization, Docker for containerized applications, and Mersk for managing app deployments and rollbacks, avoiding the complexities of Kubernetes.

How did the cost of owning their hardware compare to renting cloud services?

The cost of owning their hardware was less than a third of the projected expenses for renting cloud services, saving the company approximately $1.5 million per year.

What challenges are associated with running their own hardware?

Challenges include potential issues with scalability, the need for hiring additional staff when the infrastructure expands, and risks such as DDoS attacks.

What insights were shared about cloud service costs?

The discussion highlighted that many companies incur substantial costs from cloud services, with the previous year's cloud budget being $3.2 million, suggesting that the savings from owning hardware could be substantial.

What is the significance of the term 'serverless' in this context?

Serverless refers to cloud computing models where users manage services without dealing with server infrastructure. The transcript discusses the risks of becoming too dependent on such services, which can complicate scaling and cost management.

What was the opinion on companies being overly reliant on cloud credits?

There was a warning against viewing cloud credits as free gifts, emphasizing that they can lead to dependencies on proprietary services that may complicate future transitions away from the cloud.

What advice was given for companies considering moving out of the cloud?

Companies were encouraged to analyze their own needs, consider whether they benefit from constant scaling, and evaluate the feasibility of a cloud exit, emphasizing that the tools for migration are available and free.