Datakhi at DevLille 2026: Take Back Control of Your Data Platform
Join us at DevLille on June 12, 2026 for our talk on data sovereignty: real-world client case study, open-source stack with Garage (alternative to MinIO), PostgreSQL, Airflow and K3s, and comparison with hyperscalers.
We'll be at DevLille on June 12, 2026
Datakhi is taking the stage at DevLille 2026. Nathan and Jonathan will deliver a 45-minute talk titled "Take Back Control of Your Data Platform Against the American Giants" — a real-world feedback on building a sovereign data platform, without AWS, Azure, or GCP.
See you on Friday, June 12, 2026 at Lille Grand Palais. Full schedule on the DevLille agenda.
Why this talk?
Since the Cloud Act was passed in 2018, a legal reality has set in: U.S. authorities can legally compel AWS, Microsoft, or Google to hand over data hosted on their infrastructure — even when that data is physically stored in Europe. This stands in direct contradiction to the GDPR.
Residency (my data is in Europe) and sovereignty (my data is shielded from any foreign jurisdiction) are two different things. The gap between them is why more and more companies are starting to ask: can we do this another way?
What we'll show
A concrete case: the data platform we designed and deployed for a B2B telecom operator client. A real production infrastructure — not a POC.
The architecture is classic on paper: DataLake → DataWarehouse → Analytics. What's less classic is that it runs entirely on a private cloud, on open-source bricks, and that it went to production in 4.5 months.
The stack
The main components we'll cover:
- Garage (European alternative to MinIO) — S3-compatible object storage
- PostgreSQL — relational data warehouse, star schema
- Apache Airflow — pipeline orchestration
- K3s + Rancher — lightweight Kubernetes cluster and its management UI
- Ansible — end-to-end Infrastructure as Code
- DuckDB — embedded SQL engine for analytical workloads
For visualization: Power BI remains fully usable on the front-end, with zero cloud dependency on the back-end.
Beyond the stack: the trade-offs
The talk isn't just about the architecture. We'll also discuss what's rarely looked at: the human workload at setup and at run, the reversibility (demonstrated in production via a MinIO → Garage migration), the contractual lock-in, and the real cost compared to the Azure or AWS equivalent. We'll close on the criteria that make a private cloud a relevant choice — or not — for your context.
Who is this for?
If you're a data architect, data engineer, CTO, IT director, or just curious to see what a modern data platform looks like without a hyperscaler: this talk is for you. Intermediate level, no Kubernetes expertise required.
Want to dig in before the talk?
We've already written extensively on this approach. A few reads to set the scene:
- Building a Sovereign Data Platform: Architecture and Lessons Learned
- Sovereign Object Storage: Garage, MinIO and European Alternatives to S3
- Integrating a Sovereign Open Source Data Infrastructure: A Practical Guide
See you on June 12, 2026 in Lille. Tickets and full agenda on devlille.fr. If you'd like to chat before or after the talk, we're available — get in touch.