Containers are easy—moving your legacy system off your VM is not
Timestamps are as accurate as they can be but may be slightly off. We encourage you to listen to the full context.
Podcast Summary
In this Stack Overflow podcast episode, host Ryan Donovan sits down with Dan Ciruli, VP and General Manager of Cloud Native at Nutanix, to explore the evolving landscape of enterprise infrastructure. They dive deep into why virtual machines (VMs) remain critical in enterprise environments despite the rise of containerized applications, and how organizations can successfully integrate both technologies. (02:28)
- Main Theme: The episode focuses on bridging legacy VM-based infrastructure with modern cloud-native containers and Kubernetes, addressing why both technologies need to coexist and how to make them communicate effectively in enterprise environments.
Speakers
Ryan Donovan
Ryan Donovan is the host of the Stack Overflow podcast and editor of the Stack Overflow blog. He specializes in covering software development trends, technology discussions, and interviewing industry experts about their experiences and insights.
Dan Ciruli
Dan Ciruli is the VP and General Manager of Cloud Native at Nutanix, with over 30 years of experience in enterprise technology. He spent seven years at Google during the early development of cloud-native technologies, was a founding member of the OpenAPI initiative, served on the Istio steering committee, and worked on gRPC protocol development. (00:53)
Key Takeaways
Containers Enable Faster Innovation, Not Just Operational Efficiency
Dan emphasizes that the primary reason to adopt containers isn't just operational benefits like security or scalability—it's because they enable developers to ship software faster and innovate more rapidly. (02:28) He uses Google's early 2000s dominance over AltaVista, MapQuest, and Hotmail as an example, explaining that Google's ability to revolutionize search, maps, and email simultaneously was largely due to their early adoption of containers, which allowed them to move much faster than competitors. The key insight is that when developers can push smaller, more frequent increments of code, organizations can innovate at unprecedented speeds.
Legacy Applications Won't Disappear—Plan Accordingly
Rather than waiting for legacy systems to be rewritten, successful enterprises recognize that millions of applications written over the past 30 years will likely never be containerized. (05:13) Dan points out that major banks have tens of thousands of applications running in VMs with no plans to eliminate them, just as mainframes haven't disappeared despite decades of predictions. The practical approach is to build infrastructure that allows new containerized applications to communicate seamlessly with existing VM-based systems, rather than creating costly silos that treat them as separate environments.
Running Kubernetes in VMs Solves Multiple Infrastructure Problems
While it might seem counterintuitive, running Kubernetes within VMs actually solves several major enterprise challenges. (10:36) This approach enables consistent networking policies across both container and VM workloads, eliminates hardware silos that create resource waste, and provides the flexibility to auto-scale clusters without manual hardware provisioning. Dan explains that most hyperscaler-managed Kubernetes already runs in VMs for these exact reasons, and enterprises can achieve the same benefits on-premises.
Make On-Premises Kubernetes as Easy as Cloud Kubernetes
The reason Kubernetes adoption lags on-premises compared to cloud environments isn't technical—it's operational complexity. (22:19) Hyperscalers made it possible to spin up secure, automatically-updated Kubernetes clusters with a button click, while on-premises implementations required extensive care and feeding. Organizations need to focus on creating the same level of operational simplicity on-premises, including automated patching, security management, and developer tooling like container registries and CI/CD pipelines.
AI Will Drive Mainframe Modernization Through Code Translation
One of the most promising applications of AI in enterprise environments is translating legacy code written in languages like COBOL or Pascal into modern implementations. (19:13) Dan suggests a systematic approach: first use AI to generate comprehensive test cases from the original code to establish behavioral baselines, then use AI to translate the code itself. This approach provides the guardrails necessary for safe modernization and represents a practical application of AI that could finally enable organizations to decommission decades-old mainframe systems.
Statistics & Facts
- Dan mentions that enterprises have been writing applications to deploy on servers or VMs for thirty years, with millions of applications still running this way. (05:13) Context: This was discussed when explaining why VMs aren't going away anytime soon and why legacy modernization is such a complex challenge.
- AWS saved 4,500 years of developer work through an AI-powered Java upgrade project. (20:37) Context: This statistic was mentioned during the discussion about AI's potential for code modernization, highlighting the massive scale of efficiency gains possible with automated code transformation.
- A cloud-native company born in the last decade spends tens of millions of dollars per month in the cloud and plans to move 80% of their workload on-premises. (24:12) Context: This example was used to illustrate the trend of organizations moving workloads back on-premises after understanding the true economics of cloud computing.
