Today's Deep-Dive: Garage

Episode description

The Deep Dive episode explores Garage, an open-source distributed object storage project designed to bring enterprise-level resilience to self-hosted users and small businesses. Garage aims to provide data redundancy and availability comparable to cloud giants like Amazon and Google, but without the massive budget. It achieves this by treating data as self-contained objects with rich metadata, stored in a flat, infinitely scalable structure. A key feature is its S3 compatibility, allowing existing software designed for Amazon’s Simple Storage Service to seamlessly integrate with a Garage cluster. Developed by Doofsler, a self-hosted service provider, Garage is released under the AGPL v3 license, ensuring improvements remain open source. The system prioritizes resilience and geographical distribution, replicating data in three distinct “zones” to withstand failures in up to two locations. Garage’s engineering philosophy embraces high latency and flaky network conditions, common for smaller operators, by leveraging concepts like Amazon’s Dynamo for availability over immediate consistency and Conflict-Free Replicated Data Types (CRDTs) for reliable data merging. It also incorporates principles from Google’s Maglev for efficient traffic routing. The project boasts a low barrier to entry, requiring minimal hardware resources (1GB RAM, 16GB disk) and supporting heterogeneous hardware configurations. Garage is sustained by public funding, notably grants from the European Commission’s Next Generation Internet (NGI) initiative, supporting its development as a public good. This approach democratizes access to advanced distributed storage, offering a decentralized alternative to large, centralized cloud providers and challenging assumptions about data safety and availability.

https://garagehq.deuxfleurs.fr/

Garage is an open-source, S3-compatible object storage system designed for geo-distributed setups, aiming to provide enterprise-level data resilience to smaller operators.
It democratizes resilient infrastructure by enabling world-class data redundancy without requiring a massive budget, suitable for self-hosted users and small businesses.

Episode description

Persons