8 years of Kubernetes platform evolution: Do’s and don’ts
This session explores the technical evolution and organizational lessons learned over eight years of building and operating Kubernetes platforms at scale. Fabián Sellés Rosa, Tech Lead at Adevinta, shares what worked, what didn’t, and what teams should consider when managing Kubernetes infrastructure across regions and developer teams.
Fabián Sellés Rosa, Tech Lead at Adevinta, presents a retrospective on eight years of running Kubernetes in production. From early experimentation with self-hosted, multitenant clusters to managing 30+ clusters across four regions today, this talk outlines the strategic and technical milestones that shaped their internal platform serving over 5,000 developers and handling 400,000 requests per second. Sellés traces the journey from manual control plane management and experimental developer abstractions to migrating to a managed control plane and implementing seamless in-place upgrades. Along the way, he reflects on lessons from blue-green deployment strategies, cluster scaling, multitenancy tradeoffs, and risk management - illustrated through real-world experiences like the adoption of FiaaS, Common Platform UI, and challenges with the nginx ingress controller. Attendees will gain practical insights into platform maturity, long-term sustainability, and how Hyrum’s Law plays out in high-scale environments. This session is a candid look at the do’s and don’ts of evolving Kubernetes platforms in complex, fast-paced organizations.
8 years of Kubernetes platform evolution: Do’s and don’ts
This session explores the technical evolution and organizational lessons learned over eight years of building and operating Kubernetes platforms at scale. Fabián Sellés Rosa, Tech Lead at Adevinta, shares what worked, what didn’t, and what teams should consider when managing Kubernetes infrastructure across regions and developer teams.
Panelist

Panelist

Panelist

Moderator

Fabián Sellés Rosa
Tech Lead, Adevinta
Fabián Sellés Rosa, Tech Lead at Adevinta, presents a retrospective on eight years of running Kubernetes in production. From early experimentation with self-hosted, multitenant clusters to managing 30+ clusters across four regions today, this talk outlines the strategic and technical milestones that shaped their internal platform serving over 5,000 developers and handling 400,000 requests per second. Sellés traces the journey from manual control plane management and experimental developer abstractions to migrating to a managed control plane and implementing seamless in-place upgrades. Along the way, he reflects on lessons from blue-green deployment strategies, cluster scaling, multitenancy tradeoffs, and risk management - illustrated through real-world experiences like the adoption of FiaaS, Common Platform UI, and challenges with the nginx ingress controller. Attendees will gain practical insights into platform maturity, long-term sustainability, and how Hyrum’s Law plays out in high-scale environments. This session is a candid look at the do’s and don’ts of evolving Kubernetes platforms in complex, fast-paced organizations.
8 years of Kubernetes platform evolution: Do’s and don’ts
This session explores the technical evolution and organizational lessons learned over eight years of building and operating Kubernetes platforms at scale. Fabián Sellés Rosa, Tech Lead at Adevinta, shares what worked, what didn’t, and what teams should consider when managing Kubernetes infrastructure across regions and developer teams.
Fabián Sellés Rosa, Tech Lead at Adevinta, presents a retrospective on eight years of running Kubernetes in production. From early experimentation with self-hosted, multitenant clusters to managing 30+ clusters across four regions today, this talk outlines the strategic and technical milestones that shaped their internal platform serving over 5,000 developers and handling 400,000 requests per second. Sellés traces the journey from manual control plane management and experimental developer abstractions to migrating to a managed control plane and implementing seamless in-place upgrades. Along the way, he reflects on lessons from blue-green deployment strategies, cluster scaling, multitenancy tradeoffs, and risk management - illustrated through real-world experiences like the adoption of FiaaS, Common Platform UI, and challenges with the nginx ingress controller. Attendees will gain practical insights into platform maturity, long-term sustainability, and how Hyrum’s Law plays out in high-scale environments. This session is a candid look at the do’s and don’ts of evolving Kubernetes platforms in complex, fast-paced organizations.
8 years of Kubernetes platform evolution: Do’s and don’ts
This session explores the technical evolution and organizational lessons learned over eight years of building and operating Kubernetes platforms at scale. Fabián Sellés Rosa, Tech Lead at Adevinta, shares what worked, what didn’t, and what teams should consider when managing Kubernetes infrastructure across regions and developer teams.
Panelist

Panelist

Panelist

Host

Fabián Sellés Rosa
Tech Lead, Adevinta
Sign up now

