Mitigating Cloud Dependency with Decentralized and Local-First Architectures
Martin Kleppmann's keynote at QCon London 2026 highlighted the risks of Europe's heavy reliance on US cloud providers, advocating for architectural shifts towards technological sovereignty. The article explores three key strategies: multi-cloud commoditization, decentralized social media protocols like AT Protocol, and local-first software design, all aimed at empowering users and reducing vendor lock-in through engineering choices.
Read original on InfoQ CloudThe Challenge of Cloud Dependency
Martin Kleppmann, author of "Designing Data-Intensive Applications," identified Europe's significant dependency on US cloud providers (AWS, Azure, Google Cloud) as a critical risk. This dependency poses threats related to data sovereignty, access control, and potential service disruptions due to geopolitical events or sanctions. Mitigating this risk requires deliberate architectural decisions that shift power away from single providers and back towards users and open standards.
Strategies for Technological Sovereignty
- Multi-Cloud and Commoditization: The ability to switch cloud providers for backend services is crucial. This is achieved through commoditization via de facto standards like the S3 API for object storage, Kubernetes for deployment, Kafka protocol for streaming, and Postgres wire protocol for databases. While it increases cost and operational complexity, and may limit advanced features to the lowest common denominator, it provides essential vendor independence.
- Decentralized Social Media (AT Protocol): The AT Protocol, underlying Bluesky, is designed for "credible exit." Users retain control over their data (username, social graph, posts) in personal data servers (PDS) that anyone can run. A relay aggregates events, and AppViews build indexes. This architecture ensures that users can switch providers without data loss, fostering competition and reducing the power of a single platform.
- Local-First Software This approach combines the best of collaborative tools (like Google Sheets) with version control systems (like Git). The user's local copy of data becomes primary, reducing cloud services to mere sync and backup mechanisms. This minimizes reliance on centralized servers, simplifies provider switching, and even enables peer-to-peer synchronization, often leveraging Conflict-Free Replicated Data Types (CRDTs) for consistency. Local-first is suitable for productivity tools but not for systems managing physical resources (e.g., bank accounts) requiring a single source of truth.
Key Takeaway
The overarching theme is that through careful engineering choices – embracing commoditization, decentralization, and local-first principles – it's possible to build systems that offer greater user freedom and reduce geopolitical and vendor-lock-in risks, ultimately shifting the balance of power in favor of the user.