CNCG Ahmedabad May 2026 In-Person Meetup @ The Intellify

Kick off the meetup with a fun icebreaker, an introduction to the Cloud Native Community Ahmedabad, and a quick overview of the day's agenda.

Docker run feels like magic. It isn't. At the bottom of the stack, beneath Kubernetes, beneath containerd, beneath runc, there are just three Linux primitives: namespaces, cgroups, and a handful of syscalls. In this talk I build a working container runtime in Go from scratch. A process that thinks it's PID 1, with its own filesystem, its own hostname, and hard memory limits enforced by the kernel. No Docker. No shortcuts. Just Go and Linux doing exactly what they were built to do.

Speaker: Ajitem Sahasrabuddhe, Lead Technology Consultant @Technogise

Get to know The Intellify®, our venue partner for this meetup. Learn about the company’s journey, products and solutions, growth story, vision, and culture, along with insights into opportunities and what they’re building.

Take a short refreshment break, grab a cup of tea or coffee, and connect with fellow attendees, speakers, and community members.

By the end of this session, Attendies will have a clear understanding of how Longhorn and Rook differ and when to use each for your workloads. Also see firsthand how they handle replication, backups, and recovery in real production environments, so Attendies will know what to expect when things get complex.

We’ll also cover the common pitfalls teams run into and share practical advice on how to survive them—without getting KO’d in your cluster. Attendies will get actionable guidance for choosing the right storage solution for your applications, based on real-world lessons from production deployments.

Finally, Attendies will walk away with best practices for managing stateful workloads in Kubernetes, so your clusters stay healthy, resilient, and easier to operate.

Speaker: Shivani Rathod, DevOps developer at Motorola Solutions INC

kube-proxy has been routing packets through iptables since Kubernetes 1.0. It works until you have thousands of services and your conntrack table is full, or you need to block a single HTTP endpoint between two services, or you want to prove that traffic between nodes is actually encrypted. In this demo-heavy session, I'll build a 3-node Kubernetes cluster with Cilium as the sole CNI no kube-proxy, no iptables NAT. I'll deploy two Go microservices and progressively lock them down: L3/L4 namespace isolation, L7 HTTP-path-aware filtering that blocks specific API routes at the kernel level, and DNS-based egress policies — all enforced by eBPF programs.

Then I'll enable transparent WireGuard encryption between nodes and prove it with tcpdump on the wire. The second half goes deeper: we'll use bpftool to inspect the actual eBPF programs and maps Cilium loaded into the kernel, trace a packet through the TC hook datapath, and visualise the entire cluster through Hubble's real-time service map and flow logs. Very few introductory slides only as this will be demo and learning heavy talk.

Everyone will walk out understanding not just how to use Cilium, but how it works underneath and more importantly, how to debug Kubernetes networking at the kernel level when things go bad.

DevOps Engineer @ ViewTrade IFSC | CNCF Kubestronaut ☸️