Call for submissions: Platform Engineering meet-ups
Rootconf
Rootconf

Call for submissions: Platform Engineering meet-ups

Share your journey of building platforms that power engineering teams

All submissions

Vishal Sharma

Keep Calm and Fail Over: Engineering Chaos on CloudNative Postgres Clusters

Submitted Jun 22, 2026

Session type: Talk (30 mins) Submission made for: Bangalore meet-ups

Session Description

While Kubernetes is brilliant for stateless workloads, running stateful workloads like relational databases on it often introduces operational fragility. Many teams reach for managed cloud databases or hand-built operators, resulting in inconsistent patterns and manual failovers. In this session, we will share Nutanix’s journey of migrating our diverse PostgreSQL footprint to a unified, Kubernetes-native solution using CloudNativePG (CNPG). We will walk through the complexity gap of database challenges in Kubernetes, why we evaluated and adopted CNPG over other operators, and how its declarative state management and built-in replication capabilities helped us achieve seamless self-healing.

To validate this resilience, we battle-tested CNPG using Chaos Engineering with Chaos Mesh. We didn’t just break things and hope for the best; we employed a hypothesis-driven approach to inject network partitions, pod failures, and resource exhaustion into our primary and replica nodes. We’ll share our methodology for modeling SLAs, our chaos experiments (such as primary network partitioning and kube-apiserver disruptions), and how these tests proved CNPG’s robustness. Attendees will see how we ensured fast, clean failovers and prioritized data correctness over availability during split-brain scenarios, ultimately managing over 500 applications across 30+ clusters with confidence.

Takeaways

  • Operational Efficiency & Resilience: Discover how to efficiently run Postgres workloads in cloud-native environments by leveraging a declarative operator for automated reconciliation, day-2 operations, and self-healing.
  • Validating High Availability with Chaos Engineering: Learn how to implement a hypothesis-driven chaos testing methodology to prove built-in failover mechanisms, validate RPO/RTO thresholds, and ensure data consistency under infrastructure stress.

Target Audience

This session is highly beneficial for Platform Engineers, Site Reliability Engineers (SREs), DevOps practitioners, and Database Administrators who are looking to run or manage highly available stateful workloads (specifically PostgreSQL) on Kubernetes at scale.

Speaker Bios

Vishal Sharma is a Member of Technical Staff at Nutanix focused on distributed systems and cloud-native infrastructure. Over the past three years, he has tackled complex platform engineering challenges, including scaling observability pipelines with OpenTelemetry and building robust developer tooling and platform automation. He is currently working to ensure high availability for stateful workloads on Kubernetes using CloudNativePG.

Rutwij Nerkar is a Member of Technical Staff at Nutanix and has over 4 years of experience working with distributed systems and platforms at scale. With previous experience of building systems at scale in product companies, he’s currently tackling platform engineering challenges at Nutanix and working on building a robust Postgres experience using CloudnativePG.

All submissions

Comments

{{ gettext('Login to leave a comment') }}

{{ gettext('Post a comment…') }}
{{ gettext('New comment') }}
{{ formTitle }}

{{ errorMsg }}

{{ gettext('No comments posted yet') }}

Register

Hosted by

Rootconf
Rootconf
We care about site reliability, cloud costs, security and data privacy