The concept of distributed SQLite—using the SQLite engine as a storage layer while managing replication through external protocols—has evolved significantly since 2014. Philip O'Toole, the creator of rqlite, notes that the first open-source system to combine the Raft consensus algorithm with SQLite was released over a decade ago. This architectural choice was driven by a need for a small amount of relational storage in a clustered system where no single node was the ideal permanent host. By placing a Raft log in front of SQLite, the system ensures that every write is replicated across a cluster, preventing data loss if a single node fails.
Over the years, the technology has branched into three distinct design corners based on different consistency and availability requirements. The first corner, represented by rqlite and Canonical's dqlite, focuses on CP (Consistency and Partition Tolerance) systems. These use a leader-based consensus model where writes must be acknowledged by a majority. The second corner emerged around 2021 with the rise of streaming recovery, exemplified by Litestream. Instead of a cluster, Litestream watches the SQLite Write-Ahead Log (WAL) and streams it to cloud storage like S3, providing disaster recovery for single-node setups. The third corner involves merge-based replication, such as cr-sqlite, which uses Conflict-free Replicated Data Types (CRDTs) to allow multiple writers to diverge and merge without a central leader, catering to local-first applications.
While the underlying technology remains open-source, the market has shifted toward managed platforms. Projects like Turso (acquired by Supabase in 2026) and Cloudflare's D1 have transformed replicated SQLite from a self-deployed infrastructure component into a serverless database product. O'Toole observes that while the core idea of using SQLite as a storage engine with external replication has been reinvented by at least eight different teams, the primary distinction today is whether the user wants to run the infrastructure themselves or pay for a managed service. rqlite remains a primary choice for those seeking a fault-tolerant relational database that they can fully understand and host independently.