TL;DR
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
rqlite creator Philip O’Toole marks 12 years since he announced the open-source project, which paired SQLite with Raft consensus to replicate writes across a cluster. His retrospective traces how later projects took different approaches, from backup and recovery to multi-writer merging and managed database services. The account is a founder’s history, not an independent survey of the field.
Philip O’Toole, the creator of rqlite, has published a retrospective marking 12 years since he announced the project, which put a Raft consensus log in front of SQLite to replicate writes across a cluster. His account traces how an early open-source implementation became part of a broader field of projects that use SQLite for clustered databases, recovery, local-first applications and hosted services.
O’Toole says he announced rqlite in September 2014 and published its source code on GitHub that weekend. The original problem came from a clustered system at jut.io that needed relational storage without a clear answer to which node should hold it. Replicating SQLite through a consensus protocol offered one solution. O’Toole kept developing the project after his employer chose another approach.
In his account, rqlite’s defining choice was to make it a CP system: writes go through a leader and require acknowledgment from a majority of nodes. A minority isolated by a network partition refuses writes rather than accepting changes that might conflict with the majority. O’Toole says this design has not changed. The first version used goraft; in early 2016, he replaced it with HashiCorp’s Raft implementation, which he says rqlite still uses.
The retrospective describes a field that now includes projects with different replication models and goals. Canonical developed dqlite as an embedded store for LXD and later MicroK8s. Litestream, released by Ben Johnson in 2021, takes a different route: it streams SQLite’s write-ahead log to cloud storage for recovery, rather than using consensus to maintain a cluster. Other projects, including cr-sqlite, explored merging changes from multiple writers, while libSQL and hosted services such as Turso extended SQLite for cloud and edge use.
Three Paths Beyond Local SQLite
The account matters because “distributed SQLite” does not describe one guarantee or one product category. Consensus replication, as used by rqlite, prioritizes agreement across a cluster and can reject writes when a majority is unavailable. WAL streaming, as used by Litestream, focuses on recovering a single-node database after a failure. Multi-writer approaches such as cr-sqlite instead aim to let changes made in separate locations merge.
Those differences affect what developers can expect during outages, how quickly a service can resume, and whether local changes can be made independently. O’Toole frames the approaches as answers to different operational questions, rather than interchangeable ways of making SQLite “distributed.” That distinction can help readers evaluate whether they need continuous availability across nodes, a low-cost recovery copy, or collaboration across multiple writers.
The shift is also commercial. O’Toole describes a move from projects developers ran themselves toward managed database products offered by platform companies. That change can make SQLite-based systems easier to adopt, while leaving customers with less visibility into how a service replicates data. The retrospective does not provide a comparative technical or commercial assessment of those services.
Raspberry Pi 5 8GB for self-hosted database server
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From rqlite to Hosted Services
O’Toole says rqlite was not the only early effort to combine SQLite with replication. Expensify had used an internal Paxos-style system for years before releasing Bedrock in 2016, and ActorDB appeared around the same period as rqlite, using Raft to replicate separate SQLite databases for actors. He acknowledges both to qualify his account of the project’s place in the history.
Canonical’s dqlite, begun in 2017, shared rqlite’s broad aim but was built as a library to embed in products, rather than as a standalone server. In 2021, Litestream popularized a different approach focused on recovery from a machine failure. Fly.io later shipped LiteFS, which O’Toole describes as filesystem-level replication using leases from Consul. In 2022, cr-sqlite and libSQL added further approaches, while Cloudflare announced D1 as a managed SQLite-based database for its edge network.
These dates and characterizations come from O’Toole’s retrospective, which is written by a participant in the field. He says the source code for rqlite appeared in 2014 and that, as far as he can tell, it was the first open-source system to combine Raft and SQLite. That is his assessment; the article does not independently establish a definitive first.
Synology NAS for SQLite database backups
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Limits of One Founder’s History
The retrospective is O’Toole’s account of the field, not a third-party review. His assertion that rqlite was the first open-source Raft-and-SQLite system is qualified by “as far as I can tell,” and the material does not independently verify that claim or compare every project that may have existed.
The article also does not supply current adoption figures, reliability comparisons, performance benchmarks or a detailed account of how each hosted service handles replication. It mentions Supabase’s 2026 acquisition of Turso, but offers no transaction details. Readers should treat the history and project descriptions as the author’s perspective, not as a present-day assessment of product status or suitability.
As an affiliate, we earn on qualifying purchases.
The Next Questions for Database Teams
O’Toole’s retrospective does not announce a new rqlite release or set out a forthcoming product milestone. The next practical step for teams considering these tools is to match the replication model to their failure and consistency requirements: whether they need a consensus-backed cluster, recovery from a stored log, or mergeable writes across locations.
For the projects and providers discussed, future developments in product support, service availability and technical design are not detailed in the account. Teams evaluating them will need current documentation and direct product information rather than relying on this historical overview.
SQLite database development book distributed systems Raft
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is rqlite?
rqlite is an open-source database project that uses SQLite as its storage engine and Raft consensus to replicate writes across a cluster. O’Toole says he announced it in September 2014.
How does rqlite handle a network partition?
According to O’Toole, writes go through a leader and require acknowledgment from a majority. A minority partition that cannot reach a majority refuses writes.
Is Litestream the same kind of system as rqlite?
No. O’Toole describes Litestream as streaming SQLite’s write-ahead log to cloud storage for recovery, while rqlite replicates writes through consensus across a cluster. They address different failure and availability needs.
Does the retrospective prove rqlite was the first project of its kind?
No. O’Toole says rqlite was, “as far as I can tell,” the first open-source system to combine Raft and SQLite. That is a qualified claim by the project creator, not an independently established finding in the source material.
Source: hn
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
