KubeCon + CloudNativeCon India 2025

Is basic monitoring and logging using prometheus and ELK stack enough? Our DevOps automation platform suffered a major Kubernetes outage due to weak observability despite having the basic setup in place.

What Went Wrong?
Our "good enough" setup fell apart when it mattered most. Logs were a mess with no correlation, & tracing was half-baked, leaving us clueless during debugging. One customer’s CI/CD pipeline overwhelmed shared resources and brought everyone down. Aggressive autoscaling overloaded the control plane. Worst of all, a single tenant’s failure affected the entire cluster.

How did we fix it?
We swapped basic Prometheus for a distributed setup that could handle the scale. Workloads got isolated per tenant and team, with multi-tenancy baked in and dedicated monitoring to match. Fine-grained autoscaling & experimentation with tracing tools(Parca/Odigos). The result? A cluster that bends but doesn’t break.

If you want to avoid your own outage horror story, this talk is for you.

As organisations push to accelerate software delivery, Kubernetes has emerged as the backbone of modern platform engineering, enabling businesses to transcend traditional infrastructure and build scalable, high-performance platforms that empower developers to effortlessly deploy, manage, and scale applications.

In this session, I’ll dive into how Kubernetes is revolutionising platform engineering — turning complex infrastructure into self-service environments that drive automation, scalability, and developer efficiency. Through real-world case studies and forward-thinking strategies, I’ll demonstrate how Kubernetes helps organisations unlock unprecedented flexibility, accelerate delivery cycles, and build platforms that scale seamlessly in today’s dynamic landscape.