filter >

Photo

How We Built Cloud-Like Kubernetes on Your Hardware with Omni and Talos

Utku Özdemir

from Sidero Labs (Germany)

About speaker

Senior Software Engineer - Sidero Labs

Utku, a Senior Software Engineer at Sidero Labs, contributes to Talos Linux, a Kubernetes-centric OS. Transitioning from Java Backend Development, he immersed himself in cloud, Kubernetes, and DevOps.

About speakers company

We are the makers of Talos: - a free, open source Linux-based OS built specifically to run Kubernetes and Omni: a SaaS to provide cloud-like experience of Kubernetes on edge & your own hardware

Abstracts

specific

Many rely on managed Kubernetes offerings from cloud providers. But what if you want to run on your own hardware? We'll dive into how we at Sidero Labs architected Omni and Talos to solve this challenge, exploring our in-house COSI framework, SideroLink for secure WireGuard connection to Omni, image factory service for OS customization. You'll see how all these components work together to handle cluster lifecycle operations, with insights into architectural patterns and design decisions.


Running Kubernetes clusters on your own hardware brings unique architectural challenges when trying to achieve cloud-like operational experience. This talk shows how we approached this problem with Omni - starting with a high-level overview of how Omni, Talos, and supporting services work together as a system.

We'll examine the overall architecture: how Talos machines connect back to Omni via SideroLink (our WireGuard-based connectivity layer), how Omni coordinates with the image factory service to provide customized OS images, and how cluster operations flow through the system. Then we'll look deeper into the internals, exploring COSI - our in-house framework inspired by Kubernetes controller-runtime that provides the foundation for both Omni and Talos.

Following the architectural overview, we'll demonstrate these patterns in action by following cluster lifecycle operations: from initial machine registration, through cluster creation and scaling, to upgrades. We'll examine how state flows through the system, how components coordinate complex operations, and how we maintain consistency across infrastructure boundaries.

You'll learn about practical approaches to building management systems that span trust boundaries, patterns for coordinating operations across multiple components, and how we solved specific challenges like machine discovery, secure communication, and maintaining cluster state.

The Program Committee has not yet taken a decision on this talk

other talks of this topic

Photo
Architectures that we can use with .NET

Alexej Sommer

Capgemini

broad
Photo
The Anatomy of a Distributed JavaScript Runtime

Peter van Vliet

Masking Technology

broad
Photo
Achieving True Layered Separation with Hexagonal Architecture in Spring Boot

Adrian Kodja

softgarden e-recruiting GmbH

specific
Photo
REST or gRPC: Best practices for modern architectures

Kristina Kraljić

PIS d.o.o.

specific
Photo
DRBD Deep Dive

Philipp Reisner

LINBIT

specific
Photo
Beyond Caching: Valkey's Advanced Data Structures in Action

Viktor Vedmich

Amazon Web Services

specific
Photo
Mastering Software Design: Best Practices for Building Robust Applications

Ambesh Singh

Visionet Systems Deutschland

broad
Photo
Open Source Ecosystem for ClickHouse on Kubernetes

Denys Kondratenko

Altinity

specific
Photo
Organizational Sustainability with Platform Engineering

Lesley Cordero

The New York Times

specific
Photo
Cloud Costs with ClickHouse and OpenCost

Denys Kondratenko

Altinity

specific
Photo
ML/AI in the cloud - State of the Art in 2025

Federico Fregosi

OpsGuru

broad