Nikhil Pathak
Senior Platform Engineer, DigitalOcean
Actions
Designing Platforms based on Kubernetes
Links
Zero-Downtime Control Plane Replacement: A Kubernetes Survival Guide
Sometimes Control plane node replacement may be necessary. These situations include security incidents and compromises, compliance and regulatory requirements, infrastructure modernization, and operating system and platform updates. A significant concern is ensuring the uninterrupted operation of stateful set workloads.Replacing control plane nodes in a Kubernetes cluster presents several challenges like API server or complete cluster unavailability, risking etcd data corruption without reliable backups. Network and load balancer reconfigurations, careful timing for rolling updates, and rollback plans are also necessary. This talk details a proven approach for replacing ALL control plane nodes. We will also demonstrate how automation can be leveraged to streamline the process, enhance robustness, and guarantee zero downtime. We will provide a detailed walkthrough of how to accomplish these essential steps using a combination of shell scripts and Ansible playbooks.
Managing Millions of Storage Volumes at Scale: Inside DigitalOcean's ArgoCD Strategy
When you're managing millions of storage volumes across 12 regions, traditional deployment approaches break down. At DigitalOcean, we transformed our Storage Platform operations using ArgoCD to bring sanity to complexity.
In this talk, we'll share how DigitalOcean's Storage Platform team turned our deployment process into a GitOps-powered engine using ArgoCD. We'll take you behind the scenes of operating our Storage Kubernetes platform, StorK8s, our storage orchestration platform that powers millions of volumes across DigitalOcean's global infrastructure.
You’ll learn:
1. How we architected a single ArgoCD instance to manage 13+ clusters across 13 regions while maintaining sub-5-minute deployment times.
2. Real-world canary and blue-green deployment patterns for stateful workloads.
3. Why centralised GitOps beats federation for our use case (and when you shouldn't follow our lead)
We’ll share what worked, what didn’t, and secret ingredients that helped us scale GitOps reliably.
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top