Enterprise tech loves a good debate—and this one’s been simmering for years: Should you go serverless or double down on Kubernetes?

In this episode of Tech Council, Duncan Mapes and Jason Ehmke break down what it actually looks like to make that decision in a real-world engineering org. No hype, no buzzwords—just decades of experience navigating cloud complexity at scale.

They dig into where serverless shines, why Kubernetes is still around (and still useful), and how to stop making decisions based on ego and start making them based on outcomes. Expect strong opinions, practical advice, and a few hard truths about infrastructure no one wants to admit.

Whether you're scaling a new product or managing legacy workloads, this episode will help you better understand the trade-offs—and how to pick the right tool for your actual goals.

🎧 New episodes drop weekly. Subscribe and join the conversation as we rethink what it means to lead in enterprise tech.

Top Takeaways:

• Serverless architecture offers operational simplicity and cost efficiency.
• Kubernetes requires deep expertise and can lead to complex management challenges.
• Trade-offs exist in every architectural decision, emphasizing the need for strategic goals.
• Collaboration with partner teams is crucial for the successful integration of serverless solutions.
• Batch processing can be effectively managed using AWS services like Batch and Fargate.
• Scaling strategies should consider both horizontal and vertical scaling options.
• Cold starts in serverless can impact performance and should be managed carefully.
• Tree shaking can reduce bundle size and improve cold start times.
• Spot instances can significantly reduce costs for background processing tasks.
• Understanding AWS limitations is essential for effective serverless architecture. State management is a critical consideration in serverless applications.
• Using external data stores for state management can enhance scalability.
• Vendor lock-in is a valid concern when using cloud services.
• Portability between serverless and containerized applications is achievable with proper architecture.
• Testing should focus on code rather than recreating the entire environment locally.
• Fargate simplifies container management by reducing complexity.
• Kubernetes can add unnecessary layers of complexity in cloud environments.
• Architecting for future growth is essential for stability and reliability.
• Planning is crucial to ensure the right tool is chosen for the job.

Connect with us:

Duncan Mapes

Jason Ehmke

DevGrid.io

DevGrid on LinkedIn

DevGrid on X