Kubernetes & Cloud Native Meetup - November Edition

Servus Kubernauts!

This meetup is heading towards the Christmas break and will likely be the last one for this year. So take the chance and attend our November Edition!

18:00 - Opening and Networking
18:30 - Welcome
18:35 - Folker Schamel, CEO at LynxStep, Deploying a Cloud Native Service Mesh for Interactive Real-Time 3D Rendering with Kubernetes: A Case Study
19:15 - Tal Zwick, Software Engineer at MetalBear, on When they go high, we go low - hooking libc calls to debug Kubernetes apps
19:45 - Closing & Networking

More about the talks:
Deploying a Cloud Native Service Mesh for Interactive Real-Time 3D Rendering with Kubernetes: A Case Study
Kubernetes is a powerful platform for deploying cloud native applications. However, some applications may have special requirements that need extra attention and customization. In this session, I will share our experience of deploying a service mesh that provides server-side interactive real-time 3D rendering for web clients. I will discuss how we solved the technical issues such as exposing stateful services directly to the clients via WebSocket and WebRTC, scaling the services horizontally based on custom metrics and in coordination with user session management, and caching hot portions of terabytes of 3D data at the pod level for fast access. You will learn how to leverage Kubernetes features and best practices to deploy scalable and reliable interactive real-time applications.

When they go high, we go low - hooking libc calls to debug Kubernetes apps
Many tools exist for debugging cloud-native applications, and they mostly take one of two approaches: they either continuously sync local files to the remote container, or they connect the local machine to the cluster. mirrord takes a novel approach: when developers run their code with mirrord, it injects a dynamic library into the process, hooks libc calls, and relays some operations to be executed on the cluster instead of locally. That way, developers can run their code locally and debug it with a debugger. With all of the code’s input and output forwarded to and from the cluster, the local process “thinks” it’s running in the cluster and can be tested and debugged in cloud conditions. In this talk we will discuss the rationale behind this approach, take a short look at some of the low level Rust code that makes it possible, and see mirrord in action with a live demo.