High Availability Across Regions and Clouds Sounds Great on Paper, Until the Bill Arrives
Multi-region and multi-cloud resilience without paying for idle infrastructure. An open-source-first approach, with Pedro Cachaldora.
Presented with Pedro Cachaldora at KubeAuto Day Amsterdam 2026, 23 March 2026, Amstel Boathouse, Amsterdam.
As organizations scale globally, many teams find themselves trapped between two bad choices: accept downtime risk, or over-engineer for resilience and watch costs spiral out of control. Add vendor lock-in and proprietary tooling to the mix, and even small architectural decisions can turn into long-term financial and operational pain.
In this session, we tackle one of the most common, and least solved, problems in cloud-native platforms: how to achieve the highest possible availability across multiple regions and clouds without paying for idle infrastructure you don’t need.
We explore a practical, open-source-first approach to designing multi-region, multi-cloud architectures that balance resilience, performance, and cost. Cloud-native building blocks can be combined to reduce blast radius, avoid over-provisioning, and keep optionality across providers, without sacrificing reliability.
Through real-world scenarios and hard technical lessons, the session offers a blueprint for building systems that stay online when regions fail, scale when traffic spikes, and stay efficient when things are quiet.